A new type of signal-rich-art image for applications of data transfer, called signal-rich-art code image, is proposed. The created code image is visually similar to a preselected target image and, with a given message embedded, achieves the effect of the so-called signal-rich art. With its function similar to that of a QR code, such a type of image is produced by encoding the message into a binary bit stream, representing the bits by binary code patterns of 2 × 2 blocks, and injecting the patterns into the target image by a novel image-block luminance modulation scheme. Each signal-rich-art code image may be printed or displayed and then is recaptured using a mobile-device camera. Skillful techniques for counting the number of pattern blocks and recognition of code patterns are also proposed for message extraction from the recaptured version of the signal-rich-art code image. Good experimental results and a comparison of them with those of an existing alternative method show the feasibility and superiority of the proposed new data transfer method.

A new secure image transmission technique is proposed, which transforms automatically a given large-volume secret image into a so-called secret-fragment-visible mosaic image of the same size. The mosaic image, which looks similar to an arbitrarily selected target image and may be used as a camouflage of the secret image, is yielded by dividing the secret image into fragments and transforming their color characteristics to be those of the corresponding blocks of the target image. Skillful techniques are designed to conduct the color transformation process so that the secret image may be recovered nearly losslessly. A scheme of handling the overflows/underflows in the converted pixels' color values by recording the color differences in the untransformed color space is also proposed. The information required for recovering the secret image is embedded into the created mosaic image by a lossless data hiding scheme using a key. Good experimental results show the feasibility of the proposed method.

An image fusion method must ideally preserve both the detail of the panchromatic image and the color of the multispectral image. Existing image fusion methods incur the gamut problem of creating new colors which fall out of the RGB cube. These methods solve the problem by color clipping which yields undesirable color distortions and contrast reductions. An improved nonlinear IHS (intensity, hue, saturation; iNIHS) color space and related color transformations are proposed in this paper to solve the gamut problem without appealing to color clipping. The iNIHS space includes two halves, one being constructed from the lower half of the RGB cube by RGB to IHS transformations, and the other from the upper half of the RGB cube by CMY to IHS transformations. While incurring no out-of-gamut colors, desired intensity substitutions and additions in substitutive and additive image fusions, respectively, are all achievable, with the saturation component regulated within the maximum attainable range. Good experimental results show the feasibility of the proposed method.

A new data hiding method via collaboratively-written articles with forged revision history records on collaborative writing platforms is proposed. The hidden message is camouflaged as a stego-document consisting of a stego-article and a revision history created through a simulated process of collaborative writing. The revisions are forged using a database constructed by mining word sequences used in real cases from an English Wikipedia XML dump. Four characteristics of article revisions are identified and utilized to embed secret messages, including the author of each revision, the number of corrected word sequences, the content of the corrected word sequences, and the word sequences replacing the corrected ones. Related problems arising in utilizing these characteristics for data hiding are identified and solved skillfully, resulting in an effective multiway method for hiding secret messages into the revision history. To create more realistic revisions, Huffman coding based on the word sequence frequencies collected from Wikipedia is applied to encode the word sequences. Good experimental results show the feasibility of the proposed method.

A new tool, called signal-rich-art image, for automatic identification and data capture applications is proposed, which is created from an artistic target image for use as a carrier of a given message. The created image is visually similar to the target image, achieving the effect of so-called signal rich art. With its function similar to those of barcodes or QR codes, such a type of image is produced by fragmenting the composing characters of the message and injecting them into the target image by a novel image-block luminance modulation scheme. Skillful techniques are also proposed for message extraction from a mobile phone-captured version of the signal-rich-art image printed on paper or displayed on a screen. Good experimental results show the feasibility of the proposed method.

A dynamic thresholding method is proposed for use in the Hough transform to detect complex curves in images robustly. While determining edge pixels contributing to the peak in the Hough space for detecting a curve with noise and other errors, the proposed method can endure the errors by detecting pixels coming from an equal-width shape which is centered at the curve with a small width everywhere along the curve. This equal-width shape detection capability is accomplished by the use of a dynamic threshold for pixel selection, which is derived from the use of the first-order directional derivative of the function describing the curve. Three conventional methods are compared to show the superiority in robustness of the proposed method via experimental results, and a real-time application of the method for quick detection of lines in omni-images is also demonstrated.

A stereo vision system using two omni-cameras for 3-D vision applications is proposed, which has an automatic adaptation capability to any system setup before 3-D data computation is conducted. The adaptation, which yields the orientations and distance of the two omni-cameras, is accomplished by detecting and analyzing the horizontal lines appearing in the omni-images acquired with the cameras and a person standing in front of the cameras. Properties of line features in environments are utilized for detecting more precisely the horizontal lines that appear as conic sections in omni-images. The detection work is accomplished through the use of carefully chosen parameters and a refined Hough transform technique. The detected horizontal lines are utilized to compute the cameras' orientations and distance from which the 3-D data of space points are derived analytically. Compared with a traditional system using a pair of projective cameras with nonadjustable camera orientations and distance, the proposed system has the advantages of offering more flexibility in camera setups, better usability in wide areas, higher precision in computed 3-D data, and more convenience for nontechnical users. Good experimental results show the feasibility of the proposed system.

A new problem of automatically designing an optimal stereo vision system using two omni-directional catadioptric cameras to yield the highest 3-D data accuracy is studied. Factors of the system configuration considered in the design include camera pose, field of vision, and mirror shape. To find the optimal vision system configuration, an analytic formula is derived to model the 3-D measurement error, which takes into consideration the variations of pixel-quantization precisions and angular resolutions in images by conducting error propagation analysis in the data computation process. The formula is then used in an optimization framework to find the optimal system configurations for different shapes of system setup environments. For regular cases with rectangular cuboid-shaped 3-D measurement and camera placement areas, two fast algorithms are proposed to solve the problem, one being bisection-based and relatively slower for deriving the optimal solution, and the other faster using analytic formulas for deriving a suboptimal solution that is proved to be close to the optimal one in precision. Experimental results of simulations and real application cases show the feasibility of the proposed method.

A new space-mapping method for object location estimation, which is adaptive to camera setup changes, for use in various automation applications is proposed. The location of an object appearing in an image is estimated by mapping image coordinates of object points to corresponding real-world coordinates using a mapping table, which is constructed in two stages, with the first stage for establishing a basic table using bilinear interpolation in the camera manufacturing environment and the second for adapting the table to changes of camera heights and orientations in the application field. Analytic equations for table adaptation are derived by skillful utilization of both image formation and camera geometry properties. Good experimental results are shown to prove the feasibility of the proposed method.

A new method of combining art image generation and data hiding to enhance the camouflage effect for various information-hiding applications is proposed. First, a new type of computer art, called line-based Cubism-like image, which keeps a characteristic of the Cubism art-abstraction by prominent lines and regions from multiple viewpoints-is proposed. In the creation process with an input source image, prominent line segments in the image are detected and rearranged to form an abstract region-type art image of the Cubism flavor. Data hiding with the minimal distortion is carried out skillfully during the process of recoloring the regions in the generated art image by shifting the pixels' colors for the minimum amount of ±1 while keeping the average colors of the regions unchanged. Based on a rounding-off property in integer-valued color computation, the proposed data hiding technique is proved by theorems to be reversible, and thus useful for lossless recovery of the cover art image from the stego-image. Four security enhancement measures are also adopted to prevent hackers from extracting embedded data correctly. Experimental results show the feasibility of the proposed method.

A new blind authentication method based on the secret sharing technique with a data repair capability for grayscale document images via the use of the Portable Network Graphics (PNG) image is proposed. An authentication signal is generated for each block of a grayscale document image, which, together with the binarized block content, is transformed into several shares using the Shamir secret sharing scheme. The involved parameters are carefully chosen so that as many shares as possible are generated and embedded into an alpha channel plane. The alpha channel plane is then combined with the original grayscale image to form a PNG image. During the embedding process, the computed share values are mapped into a range of alpha channel values near their maximum value of 255 to yield a transparent stego-image with a disguise effect. In the process of image authentication, an image block is marked as tampered if the authentication signal computed from the current block content does not match that extracted from the shares embedded in the alpha channel plane. Data repairing is then applied to each tampered block by a reverse Shamir scheme after collecting two shares from unmarked blocks. Measures for protecting the security of the data hidden in the alpha channel are also proposed. Good experimental results prove the effectiveness of the proposed method for real applications.

A new type of computer art image called secret-fragment-visible mosaic image is proposed, which is created automatically by composing small fragments of a given image to become a target image in a mosaic form, achieving an effect of embedding the given image visibly but secretly in the resulting mosaic image. This effect of information hiding is useful for covert communication or secure keeping of secret images. To create a mosaic image of this type from a given secret color image, the 3-D color space is transformed into a new 1-D colorscale, based on which a new image similarity measure is proposed for selecting from a database a target image that is the most similar to the given secret image. A fast greedy search algorithm is proposed to find a similar tile image in the secret image to fit into each block in the target image. The information of the tile image fitting sequence is embedded into randomly-selected pixels in the created mosaic image by a lossless LSB replacement scheme using a secret key; without the key, the secret image cannot be recovered. The proposed method, originally designed for dealing with color images, is also extended to create grayscale mosaic images which are useful for hiding text-type grayscale document images. An additional measure to enhance the embedded data security is also proposed. Good experimental results show the feasibility of the proposed method.

A new approach to efficient authentication of quotations from trusted sources but embedded in messages created by unknown authors is proposed. Two different techniques for efficient generation of authenticable quotations using cascaded hashing are proposed for conventional quotations, and an efficient two-dimensional signature generation technique is proposed for two-dimensional quotations from spreadsheets. A technique for associating the signature with the quotation in the Microsoft Word document is also proposed for generation of integrated authenticable quotations that survive copy-and-paste operations.

A novel method for generic visible watermarking with a capability of lossless image recovery is proposed. The method is based on the use of deterministic one-to-one compound mappings of image pixel values for overlaying a variety of visible watermarks of arbitrary sizes on cover images. The compound mappings are proved to be reversible, which allows for lossless recovery of original images from watermarked images. The mappings may be adjusted to yield pixel values close to those of desired visible watermarks. Different types of visible watermarks, including opaque monochrome and translucent full color ones, are embedded as applications of the proposed generic approach. A two-fold monotonically increasing compound mapping is created and proved to yield more distinctive visible watermarks in the watermarked image. Security protection measures by parameter and mapping randomizations have also been proposed to deter attackers from illicit image recoveries. Experimental results demonstrating the effectiveness of the proposed approach are also included.

A new steganographic method for data hiding in Microsoft Word documents by a change tracking technique is proposed. The data embedding is disguised such that the stegodocument appears to be the product of a collaborative writing effort. Text segments in the document are degenerated, mimicking to be the work of an author with inferior writing skills, with the secret message embedded in the choices of degenerations. The degenerations are then revised with the changes being tracked, making it appear as if a cautious author is correcting the mistakes. The change tracking information contained in the stegodocument allows the original cover, the degenerated document, and, hence, the secret message to be recovered. The extra change tracking information added during message embedding is vital in a normal collaboration scenario, and so hinders ignorant removals by skeptics. Experiments demonstrate the feasibility of the proposed method

A unified approach to unwarping of omni-images into panoramic or perspective-view images is proposed. The approach does not adopt the conventional technique of calibrating the related parameters of an omni-camera. Instead, it is based on a new concept of pano-mapping table, which is created once forever by a simple learning process for an omni-camera of any kind as a summary of the information conveyed by all the camera parameters. The learning process takes as input a set of landmark point pairs in the world space and in a given image. With the help of the pano-mapping table, any panoramic or perspective-view image can be created from an input omni-image taken by the omni-camera according to an analytic computation process proposed in this study. Experimental results show the feasibility of the proposed approach.

A novel block-based binary image authentication technique is proposed. The authentication signals of each 9/spl times/9 image block are placed in a selected 3/spl times/3 block within the 9/spl times/9 block. The authentication signal contains a certain relationship contributed by the standard deviation values of the other eight 3/spl times/3 blocks. Authentication signals are generated by rearranging the pixels in the selected 3/spl times/3 block. The rearrangement is designed according to a property of inverse halftoning to keep the visual change in the selected block imperceptible. By comparing the arrangement of the block pixels in a suspicious image with the extracted authentication signals, the fidelity and integrity of the image can be verified.

Palette images are widely used in multimedia and Internet applications. In this paper, a new method for data hiding in palette images with security protection by color ordering and mapping, as well as parameter randomization, is proposed. First, image pixels are classified as data embeddable or nonembeddable, and only the former ones are used to embed secret data. The proposed idea of data hiding is based on the use of a new type of color-ordering relationship, from which a color-mapping function is defined with binary values as output. When a secret data bit is to be embedded, a data-embeddable pixel is selected, and its color is adjusted to make the output of the color-mapping function equal to the secret bit value. The embedded secret data can be extracted correctly and quickly from the resulting stego-image by merely inspecting the outputs of the color-mapping function. Indetectability of the secret information embedded by the proposed method is also analyzed and confirmed. Furthermore, a number of possible security enhancement measures based on parameter randomization in the data-embedding process are proposed to protect the hidden data in the stego-image. The randomization effect is created by the use of a secret key and a number of random number-generating functions. The proposed data hiding method was tested with a variety of palette images. The experimental results show that secret data can be embedded and extracted successfully without producing visual artifacts in the cover image. A good balance between stego-image quality and data-embedding capacity can be achieved, which proves the efficiency and feasibility of the proposed method for practical applications.

A novel approach to location estimation and trajectory prediction of moving lateral vehicle locations for driving assistance using wheels shape information in single 2-D vehicle images by 3-D computer vision techniques is proposed. The location scheme is supposed to be performed on a vehicle with a camera mounted on the front bumper. An analytical solution is applied to estimate the location of the lateral vehicle. Firstly, the normal vectors of the car wheels in the camera coordinate system as well as the projections of the wheel centers in the image plane are derived. Then orientation angle of the lateral vehicle with respect to the driving vehicle is derived from the two vectors. Next, the image projections of the contact points of the wheels with the ground are detected. Finally, the image positions of the contact points are used to determine the 3D relative position of the moving vehicle with respect to the camera coordinate system by the back-projection principle. After the location of a moving lateral vehicle is obtained, a new approach is proposed to predict the trajectory of a lateral vehicle by using the spatial information of the two wheels in a single image. Real images are tested and experimental results show the effectiveness of the proposed approaches. The results are useful for car driving assistance and vehicle collision avoidance.

A vision-based system for tracking and interpreting leg motion in image sequences using a single camera is developed for a user to control his movement in the virtual world by his legs. Twelve control commands are defined. The trajectories of the color marks placed on the shoes of the user are used to determine the types of leg movement by a first-order Markov process. Then, the types of leg movement are encoded symbolically as input to Mealy machines to recognize the control command associated with a sequence of leg movements. The proposed system is implemented on a commercial PC without any special hardware. Because the transition functions of Mealy machines are deterministic, the implementation of the proposed system is simple and the response time of the system is short. Experimental results with a 14-Hz frame rate on 320/spl times/240 image resolution are included to prove the feasibility of the proposed approach.

To develop a reliable computer vision system, the employed algorithm must guarantee good output quality. In this study, to ensure the quality of the pose estimated from line features, two simple test functions based on statistical hypothesis testing are defined. First, an error function based on the relation between the line features and some quality thresholds is defined. By using the first test function defined by a lower bound of the error function, poor input can be detected before estimating the pose. After pose estimation, the second test function can be used to decide if the estimated result is sufficiently accurate. Experimental results show that the first test function can detect input with low qualities or erroneous line correspondences and that the overall proposed method yields reliable estimated results.

A vision-based approach to obstacle avoidance for autonomous land vehicle (ALV) navigation in indoor environments is proposed. The approach is based on the use of a pattern recognition scheme, the quadratic classifier, to find collision-free paths in unknown indoor corridor environments. Obstacles treated in this study include the walls of the corridor and the objects that appear in the way of ALV navigation in the corridor. Detected obstacles as well as the two sides of the ALV body are considered as patterns. A systematic method for separating these patterns into two classes is proposed. The two pattern classes are used as the input data to design a quadratic classifier. Finally, the two-dimensional decision boundary of the classifier, which goes through the middle point between the two front vehicle wheels, is taken as a local collision-free path. This approach is implemented on a real ALV and successful navigations confirm the feasibility of the approach.

An incremental learning by navigation approach to vision based autonomous land vehicle (ALV) guidance in indoor environments is proposed. The approach consists of three stages: initial learning, navigation, and model updating. In the initial learning stage, the ALV is driven manually, and environment images and other status data are recorded automatically. Then, an offline procedure is performed to build an initial environment model. In the navigation stage, the ALV moves along the learned environment automatically, locates itself by model matching, and records necessary information for model updating. In the model updating stage, an offline procedure is performed to refine the learned model. A more precise model is obtained after each navigation-and-update iteration. Used environment features are vertical straight lines in camera views. A multiweighted generalized Hough transform is proposed for model matching. A real ALV was used as the testbed, and successful navigation experiments show the feasibility of the proposed approach.

A new color image compression technique based on moment-preserving and block truncation coding is proposed. An input image is divided into nonoverlapping blocks and each block pixel is assigned one of two representative colors, which are computed with analytic formulas derived from preserving certain moments in the block. A bit map is then generated for each block to represent the pixels' colors. Different uniformity conditions in the representative colors are also identified and utilized to save code bits. Good average compression ratios up to about 13 can be achieved, as shown by experimental results.

A new color image compression technique based on moment-preserving and block truncation coding (BTC) is proposed. An input image is divided into non-overlapping blocks. And each block is coded by two representative colors, or equivalently, by two sets of tristimulus values (R/sub 1/, G/sub 1/, B/sub 1/) and (R/sub 2/, G/sub 2/, B/sub 2/). To compute the two representative colors of each block, some color moments of the block are preserved, and the two sets of desired tristimulus values are computed accordingly using analytic formulas. The compression ratio is 6 when the block size is 4/spl times/4. Furthermore, a simple method is applied to quantize the bit map into 64 predefined bit maps and requantize each color component from 8 bits into 4 bits. An average compression ratio of 14.00 can be achieved.>

We propose a new method to perform two-class clustering of 2-D data in a quick and automatic way by preserving certain features of the input data. The method is analytical, deterministic, unsupervised, automatic, and noniterative. The computation time is of order n if the data size is n, and hence much faster than any other method which requires the computation of an n-by-n dissimilarity matrix. Furthermore, the proposed method does not have the trouble of guessing initial values. This new approach is thus more suitable for fast automatic hierarchical clustering or any other fields requiring fast automatic two-class clustering of 2-D data. The method can be extended to cluster data in higher dimensional space. A 3-D example is included.>

Consecutive strokes of Chinese characters tend to be connected in fast writing, and this causes a problem for most stroke-based recognition approaches. A recognition scheme for recognizing cursive Chinese characters under the constraint of correct stroke writing orders is proposed. The recognition scheme consists of two phases: candidate character selection and detailed matching. In the former phase, an input script with N strokes is used to split the strokes of each reference character into N corresponding parts. In the latter phase, the connected input strokes are broken into multiple strokes under the guidance of candidate characters. In both phases, dynamic programming is employed for stroke or character matching. Good experimental results prove the feasibility of the proposed approach for cursive Chinese character recognition.>

A neural-network implementation of the moment-preserving technique, which is widely used for image processing, is proposed. The moment-preserving technique can be thought of as an information transformation method which groups the pixels of an image into classes. The variables in the so-called moment-preserving equations are determined iteratively by a recurrent neural network and a connectionist neural network which work cooperatively. Both of the networks are designed in such a way that the sum of square errors between the moments of the input image and those of the output version is minimized. The proposed neural network system is applied to automatic threshold selection. The experimental results show that the system can threshold images successfully. The performance of the method is compared with those of four other histogram-based multilevel threshold selection methods. The simulation results show that the proposed technique is at least as good as the other methods.>

A single-layer recurrent neural network is proposed to perform thinning of binary images. This network iteratively removes the contour points of an object shape by template matching. The set of templates is specially designed for a one-pass parallel thinning algorithm. The proposed neural network produce the same results as the algorithm. Neurons in the neural network performs a sigma-pi function to collect inputs. To obtain this function, the templates used in the algorithm are transformed to equivalent Boolean expressions. After the neural network converges, a perfectly 8-connected skeleton is derived. Good experimental results show the feasibility of the proposed approach.

The problem of handwritten Chinese character recognition is solved by matching character stroke segments using an iteration scheme. Length and orientation similarity properties, and coordinate overlapping ratios are used to define a measure of similarity between any two stroke segments. The initial measures of similarity between the stroke segments of the input and template characters are used to set up a match network which includes all the match relationships between the input and template stroke segments. Based on the concept of at-most-one to one mapping an iteration scheme is employed to adjust the match relationships, using the contextual information implicitly contained in the match network, so that the match relationships can get into a stable state. From the final match relationships, matched stroke-segment pairs are determined by a mutually-best match strategy and the degree of similarity between the input and each template character is evaluated accordingly. Certain structure information of Chinese characters is also used in the evaluation process. The experimental results show that the proposed approach is effective. For recognition of Chinese characters written by a specific person, the recognition rate is about 96%. If the characters of the first three ranks are checked in counting the recognition rate, the rate rises to 99.6%.

A novel approach to camera calibration by vanishing lines is proposed. Calibrated parameters include the orientation, position, and focal length of a camera. A hexagon is used as the calibration target to generate a vanishing line of the ground plane from its projected image. It is shown that the vanishing line includes useful geometric hints about the camera orientation parameters and the focal length, from which the orientation parameters can be solved easily and analytically. And the camera position parameters can be calibrated by the use of related geometric projective relationships. The simplicity of the target eliminates the complexity of the environment setup and simplifies the feature extraction in relevant image processing. The calibration formulas are also simple to compute. Experimental results show the feasibility of the proposed approach.>

A novel approach to the determination of robot locations by common object shapes is proposed. Any object that has a polygon-shaped top and a lateral surface perpendicular to the top can be used for robot locations in the approach, as long as the top shape is known in advance. In addition, the solution provided by the approach can be computed analytically. These merits make the proposed approach more practical for general applications than other approaches using specially designed marks or requiring iterative computation. From a monocular image of an object, image processing and numerical analysis techniques are applied to extract the projection characteristics of the polygon corners on the object top surface, from which the position and the orientation parameters of a camera-mounted robot can be determined. Experimental results with location errors less than 5% prove the feasibility of the proposed approach. Error analysis that is useful for choosing better viewing angles to get more accurate location results is included.>

The fact that there exists a large number of homonyms in Chinese characters is a problem for the phonetic input method which is favored by most casual users for Chinese input into computers. A new approach to solving the problem is proposed. The input method is regarded as a process of assigning syllables to words. The proposed approach makes use of the fact that little or even no ambiguity exists in determining the characters in a multi-syllabic word. The formation relationships among the words in an input sentence are utilized to perform a relaxation-based word identification process. An advantage of the approach is that once the composing words of the sentence are identified, the desired characters are also correctly selected. Experiments have been conducted with a hit rate of over 96%, thus showing the feasibility of this approach.

Due to noise and distortion, segmentation uncertainty is a key problem in structural pattern analysis. In this paper we propose the use of the split operation for shape recognition by attributed string matching. After illustrating the disadvantage of attributed string matching using the merge operation, the split operation is proposed. Under the guidance of the model shape, an input shape can be reapproximated, using the split operation, into a new attributed string representation. By combining the split and the merge operations for shape matching it is unnecessary to apply any type of edit operation to a model shape. This makes the distance between the input shape and the model shape more meaningful and stable, and improves recognition results. An algorithm for attributed string matching by split-and-merge is proposed. To eliminate the effect of the numbers of primitives in the model shape on the shape distance, shape recognition based on a similarity measure is also proposed. Good experimental results prove the feasibility of the proposed approach for general shape recognition.

A novel method for curve detection based on the moment-preserving principle is proposed. The method can be used to estimate curve locations and widths to subpixel accuracy. For each 4.5-unit circle in an input image that includes a curve segment, the approach derives a parabolic equation as well as a width value to describe the curve segment. Experimental results are included to show the effectiveness of the proposed detector.>

A new approach to 3-D object inspection using 2-D camera views is proposed. With input objects put on a turntable for inspection, a top view of the object is first taken by a top camera right above the table. The object shape boundary points are extracted, and the centroid as well as the principal axis of the shape are computed for top-view registration with the model shape, using the generalized Hough transform. The distance-weighted correlation is then used as the similarity measure for shape matching. Input objects with low similarity values are rejected. Non-rejected objects are inspected further from lateral views, using a lateral camera for image taking and the turntable for object rotation. Experimental results with a high inspection rate and a fast inspection speed show the feasibility of the proposed approach.

The objective of this paper is to develop an object recognitlon system through the combination of 2-D tactile image array and visual sensors. A video camera is used to acquire a top view image of an object and two tactile sensing arrays mounted on a gripper are used to detect the tactile information about the lateral surfaces of the object. 3-D reference object models are established as a decision tree, and recognition of unknown objects is accomplished through measuring and comparing input object features hierarchically with these of the reference objects associated with the decision tree. The clustering process and recognition procedures are described. The recognition scheme has been implemented. The resulting decision tree is also presented.

Image analysis techniques are applied to adaptive automatic vehicle navigation. The proposed image-based navigation system is made adaptive to follow any selected path embedded in a curve-type path network. This is achieved with three major capabilities of the proposed system: path network learning, reference path setup, and guided path navigation. The first capability enables the system to extract relevant information out of a given network map, and the second collects along-path reference data for a selected path from the extracted network information. During guided path navigation, consecutive path images are taken by a television camera on the vehicle and then analyzed for navigation control along path curves and for angular turning at path crossings. The control structure of the automatic navigation process is modelled as a Moore-type sequential machine in automata theory. Correct path navigation is ascertained by verifying each path crossing encountered on the road against the reference data by an image matching technique. Simulation of vehicle movement with a computer-controlled pantilt is also described. Simulation results show the feasibility of the proposed approach.

A new structural approach to shape recognition using attributed string matching with merging is proposed. After illustrating the disadvantages of conventional symbolic string matching using changes, deletions, and insertions, attributed strings are suggested for matching. Each attributed string is an ordered sequence of shape boundary primitives, each representing a basic boundary structural unit, line segment, with two types of numerical attributes, length and direction. A new type of primitive edit operation, called merge, is then introduced, which can be used to combine and then match any number of consecutive boundary primitives in one shape with those in another. The resulting attributed string matching with merging approach is shown useful for recognizing distorted shapes. Experimental results prove the feasibility of the proposed approach for general shape recognition. Some possible extensions of the approach are also included.

A graph matching approach is proposed in this paper for solving the task assignment problem encountered in distributed computing systems. A cost function defined in terms of a single unit, time, is proposed for evaluating the effectiveness of task assignment. This cost function represents the maximum time for a task to complete module execution and communication in all the processors. A new optimization criterion, called the minimax criterion, is also proposed, based on which both minimization of interprocessor communication and balance of processor loading can be achieved. The proposed approach allows various system constraints to be included for consideration. With the proposed cost function and the minimax criterion, optimal task assignment is defined. Graphs are then used to represent the module relationship of a given task and the processor structure of a distributed computing system. Module assignment to system processors is transformed into a type of graph matching, called weak homomorphism. The search of optimal weak homomorphism corresponding to optimal task assignment is next formulated as a state-space search problem. It is then solved by the well-known A* algorithm in artificial intelligence after proper heuristic information for speeding up the search is suggested. An illustrative example and some experimental results are also included to show the effectiveness of the heuristic search.

A vision system for automated parts inspection is proposed. The system is equipped with learning capabilities such that it automatically selects from a set of sample parts a minimum, but effective inspection region within the camera's field of view for parts discrimination. A binary template is formed within the inspection region which is then used for parts inspection by template matching. The inspection speed is enhanced by keeping the inspection region small and by making the matching task uncomplicated. A simple learning algorithm based on statistical pattern recognition theory is employed, which only requires the system to be taught by a training set of good and defective parts without specific defect identification or location. The system is applicable to most 2-D industrial parts inspection.

The structure-preserved error-correcting graph isomorphism proposed by Tsai and Fu (1979) for matching patterns represented by attributed relational graphs is extended to the case of subgraphs. The resulting subgraph error-correcting isomorphism, which includes the structure-preserved error-correcting graph isomorphism as a special case, is useful for recognizing partially viewed or structurally distorted patterns. After formulating a subgraph error-correcting isomorphism as a state-space tree-search problem, heuristic information useful for speeding up the search is suggested and an ordered-search algorithm is proposed for finding an optimal subgraph error-correcting isomorphism.

Attributed grammars are defned from the pattern recognidon point of view and shown to be useful for descriptions of syntactic stuctures as well as semantic attributes in primitives, subpatterns, and patterns. A pattern analysis system using attributed grammars Is proposed for pattern classification and description. This system extracts primitives and their attributes after preprocessing, performs syntax analysis of the resulting pattern representations, computes and extracts subpattern attributes for syntactically accepted patterns, and finally makes decisions according to the Bayes decision rule. Such a system uses a combination of syntactic and statistical pattern recognition techniques, as is demonstrated by illustrative examples and experimental results.

A pattern deformational model is proposed in this paper. Pattern deformations are categorized into two types: local deformation and structural deformation. A structure-preserving local deformation can be decomposed into a syntactic deformation followed by a semantic deformation, the former being induced on primitive structures and the latter on primitive properties. Bayes error-correcting parsing algorithms are proposed accordingly which not only can perform normal syntax analysis but also can make statistical decisions. An optimum Bayes error-correcting recognition system is then formulated for pattern classification. The system can be considered as a hybrid pattern classifier which uses both syntactic and statistical pattern recognition techniques.

The pattern deformational model proposed by Tsai and Fu [11] is extended so that numerical attributes and probability or density distributions can be introduced into primitives and relations in a nonhierarchical relational graph. Conventional graph isomorphisms are then generalized to include error-correcting capability for matching deformed patterns represented by such attributed relational graphs. An ordered-search algorithm is proposed for determining error-correcting isomorphisms. Finally, a pattern classification approach using graph isomorphisms is described, which can be considered as a combination of structural and statistical techniques.