This paper proposes number-driven perceptual segmentation of natural color images using a fuzzy-based hierarchical algorithm for an easy decision of the optimal segmentation result. A fuzzy-based homogeneity measure makes a fusion of the L*a*b* color features and the SGF texture features. Proposed hierarchical segmentation method is performed in four stages: simple splitting, local merging, global merging and boundary refinement. The effectiveness of the proposed method is confirmed through computer simulations that demonstrate an easy determination of the optimal segmentation result.

Telecommunication networks are spreading worldwide. People can communicate with each other beyond spacial restrictions using the Internet. In this situation, the demand for transmission bandwidth and storage space continues to outstrip the capacity of existing technologies. Image and video compression technology is therefore essential for the effective use of communication networks. In this paper, we propose a new method to compress images using vector quantization. Dimensions of the training vectors to prepare a codebook are determined on the basis of local fractal dimensions (LFDs) of a learning image. This means that each block size to divide the learning image is specified by the LFDs. In principle, the smaller the number of pixels in a training vector is, the larger is a codebook (low bit rate). Furthermore, the smaller the number of pixels in a codebook is, the higher is the quality of the encoded image. This is a tradeoff between compression rate and quality of the encoded image. The tradeoff can be solved by the division of images with different block sizes. Furthermore, the code-vectors are computed from the training vectors using a fuzzy k-means clustering algorithm. The performance of the proposed algorithm was evaluated by compression rate and quality of encoded images in comparison with those using fuzzy generalized Lloyd algorithm. The experiments showed that the proposed algorithm solved the tradeoff between compression rate and quality of image.

These days, accessibility-related regulations and guidelines have been accelerating the improvement of Web accessibility. One of the accelerating factors is the development and deployment of accessibility evaluation tools for authoring time and repair time. They mainly focus on creating compliant Web sites by analyzing the HTML syntax of pages, and report that pages are compliant when there are no syntactical errors. However, such compliant pages are often not truly usable by blind users. This is because current evaluation tools merely check if the HTML tags are appropriately used to be compliant with regulations and guidelines. It would be better if such tools paid more attention to real usability, especially on time-oriented usability factors, such as the speed to reach target content, the ease of understanding the page structure, and the navigability, in order to help Web designers to create not simply compliant pages but also usable pages for the blind. Therefore, we decided to develop Accessibility Designer (aDesigner), which has capabilities to visualize blind users' usability by using colors and gradations. The visualization function allows Web designers to grasp the weak points in their pages, and to recognize how accessible or inaccessible their pages are at a glance. In this paper, after reviewing the related work, we describe our approach to visualize blind users' usability followed by an overview of Accessibility Designer. We then report on our evaluations of real Web sites using Accessibility Designer. After discussing the results, we conclude the paper.

The Web has become a new information resource for the blind. However, Web accessibility is becoming worse, since page authors tend to care only for the visual appearance. We have developed an Accessibility Transcoding System to solve this problem. This system has the ability to transcode complete pages on annotated sites into totally accessible pages without changing the original pages. However, site-wide annotation authoring is an extremely tedious and time-consuming task. This prevented us from applying our transcoding system to a wide variety of sites. In order to overcome this difficulty, we developed a new algorithm, "Dynamic Annotation Matching". By utilizing this algorithm, our transcoding system can automatically determine appropriate annotations based on each page's layout. We also developed a site-wide annotation-authoring tool, "Site Pattern Analyzer." We evaluated the feasibility of creating site-wide annotations by using the algorithm and the tool, and report on our success here.

We describe the overall role of soft computing (SC) in signal processing and pattern recognition (SPPR) with specific applications to biomedical engineering, geoscience for mining and civil engineering human interfaces, and image processing. Detection of characteristic points in an electrocardiogram to implement an advanced ECG analyzer is presented which is carried out using both conventional SPPR techniques and self-organizing neural networks. Successful technologies for monitoring a geostructure by supervised and self-organizing neural networks are described. Identification of a freehand drawing by a combination of fuzzy logic and neural networks is also described. Moreover, application of fuzzy logic to image segmentation is presented. Finally, innovation of SPPR using SC technologies is discussed.

Presents a new automatic circular decomposition algorithm, which proceeds the separation of connected circular particles, in order to locate their center coordinates and estimate their radii. This new algorithm is based on the following supposition "if you are looking for circles you must assume that all objects in an image are circles, until you can prove they are not". In this work the authors compare this heuristic algorithm with the polygonal approximation based algorithm proposed by Kubo (1988) in their methods and results. Both of them are able to decompose connected blood cells with some differences: the proposed algorithm is implemented in 2 steps, while the polygonal approximation based method is divided in 4 steps and its input parameters present high sensitivity to noise.

We present a rough and an accurate segmentation of natural color images using a fuzzy region-growing algorithm. In the proposed method, the color difference and the color gradient are used as the pixel features to produce an accurate segmentation, while the local fractal dimension is used as the region feature to yield a rough segmentation in a natural color image. The effectiveness of the proposed method is confirmed through computer simulations that demonstrate a rough segmentation at the fine-texture regions and an accurate segmentation at the strong-edge regions simultaneously.

Existing fractal methods of texture analysis rely on the fractal dimension of textures as a function of scale for their discrimination and classification. We propose a method which is based on the possible multiscaling/multifractality of textures. A stochastic model is suggested to represent this multiscaling behaviour. We demonstrate the value of the method on a number of similar data sets (hence quite difficult for their discrimination) from the high-resolution Brodatz album.

In this paper we present a method for representing and retrieving video sequences based on the domain-specific behavior of objects present in the scenes. The representation includes three parts: (a) Action description, representing the action performed by a single object. (b) Interaction description, which describes interactions between multiple objects and is mapped directly to the event semantics in the content domain. (c) Event Structure, which provides a set of spatial and temporal relationship functions, along with a syntax to define the necessary conditions that a particular interaction should meet. Retrieval is performed by processing Event Structures, interpreting object relationships and selecting the relevant combinations of Action descriptions which match the conditions defined in the Event Structures. We describe an implementation of this system for retrieving scenes of soccer plays from among several soccer video sequences.

We present a rough and an accurate segmentation of natural images using a fuzzy region-growing algorithm. First, an optimum number of the blanket for local areas is determined to estimate the optimal local fractal dimension. Then, the intensity features and the local fractal-dimension feature are integrated into the fuzzy region-growing algorithm. In the proposed method, the intensity features are used to produce an accurate segmentation, while the fractal-dimension feature is used to yield a rough segmentation in a natural image. The effectiveness of the proposed method is confirmed through computer simulations that demonstrate a rough segmentation at the fine-texture regions and an accurate segmentation at the strong-edge regions simultaneously.

We propose a multiwindowed approach to the estimation of a local fractal-dimension map as an image feature for a natural image segmentation. The fractal dimension is estimated by the blanket method by using the optimum number of blankets. The local fractal-dimension map is created for a natural image using several sizes of windows which are selected by fuzzy set theory. We also propose an image segmentation algorithm that combines a rough and an accurate segmentation by integrating different features into the region growing algorithm. The effectiveness of the multiwindowed approach and the proposed segmentation algorithm are confirmed through computer simulations.

We present a new method that integrates intensity features and a local fractal-dimension feature into a region growing algorithm for the segmentation of natural images. A fuzzy rule is used to integrate different type of feature into a segmentation algorithm. In the proposed algorithm, intensity features are used to produce an accurate segmentation, while the fractal-dimension feature is used to yield a rough segmentation in a natural image. The effective combination of the different features provides the segmented results similar to the ones by a human visual system.

An associative memory (AM) system using fuzzy sets is presented in this paper. Vectors to be stored in the AM are transformed into fuzzy sets. Fuzzy rules are constructed from the fuzzified vectors to associate vector when a stimulus vector is input to the AM system. Computational experiments are carried out to evaluate performance of the AM system. The experiments show that the AM system is responsive to feature of the stimulus vectors. In addition, the AM system performs more than 95% correct association rate until 20% of the binary valued elements of the stored vectors are flipped randomly from either zero or one.

We present a method that integrates local fractal dimension and edge information into a region growing algorithm for the segmentation of natural images. We compare two methods of estimating the local fractal dimension in the proposed segmentation algorithm. One is a blanket method and the other is a Fourier-wavelet method. We also propose a technique to store the edge information not on a pixel itself but on a boundary between pixels in the region-edge integrating algorithm in order to use the edge information more effectively and to simplify the algorithm.

In robot vision the detection of edges of an image serves to simplify the data, in the early stages, into a form which is more easily processed. But because of noise or due to the inherent weaknesses of the chosen edge detector, gaps or interruptions in the edges may occur. These gaps must be filled or linked to form a more continuous edge. A new method of edge linking is proposed that makes use of a weighting system based on edge segment position, direction, and magnitude in deciding the optimal choice of which edge segments to link. Edge pixels are labeled and formed into edge segments, and the edge segments are then linked based on the decision of the weighting system. As a result of the unique way in which the gaps between edge segments are filled, distortion of the edge image is kept minimal, the proposed edge linker can be combined with different edge detection schemes. Comparisons with a method previously proposed by the authors shows an improvement in the quality of the linked edge map produced as well as a savings in computational time.