The problem of learning a proper distance or similarity metric arises in many applications such as content-based image retrieval. In this work, we propose a boosting algorithm, MetricBoost, to learn the distance metric that preserves the proximity relationships among object triplets: object i is more similar to object j than to object k. Metric-Boost constructs a positive semi-definite (PSD) matrix that parameterizes the distance metric by combining rank-one PSD matrices. Different options of weak models and combination coefficients are derived. Unlike existing proximity preserving metric learning which is generally not scalable, MetricBoost employs a bipartite strategy to dramatically reduce computation cost by decomposing proximity relationships over triplets into pair-wise constraints. Met-ricBoost outperforms the state-of-the-art on two real-world medical problems: 1. identifying and quantifying diffuse lung diseases; 2. colorectal polyp matching between different views, as well as on other benchmark datasets.

3D object detection and importance regression/ranking are at the core for semantically interpreting 3D medical images of computer aided diagnosis (CAD). In this paper, we propose effective image segmentation features and a novel multiple instance regression method for solving the above challenges. We perform supervised learning based segmentation algorithm on numerous lesion candidates (as 3D VOIs: Volumes Of Interest in CT images) which can be true or false. By assessing the statistical properties in the joint space of segmentation output (e.g., a 3D class-specific probability map or cloud), and original image appearance, 57 descriptive features in six subgroups are derived. The new feature set shows excellent performance on effectively classifying ambiguous positive and negative VOIs, for our CAD system of detecting colonic polyps using CT images. The proposed regression model on our segmentation derived features behaves as a robust object (polyp) size/importance estimator and ranking module with high reliability, which is critical for automatic clinical reporting and cancer staging. Extensive evaluation is executed on a large clinical dataset of 770 CT scans from 12 medical sites for validation, with the best state-of-the-art results.

Accurate and automatic colonic polyp segmentation and measurement in Computed Tomography (CT) has significant importance for 3D polyp detection, classification, and more generally computer aided diagnosis of colon cancers. In this paper, we propose a three-staged probabilistic binary classification approach for automatically segmenting polyp voxels from their surrounding tissues in CT. Our system integrates low-, and mid-level information for discriminative learning under local polar coordinates which align on the 3D colon surface around detected polyp. More importantly, our supervised learning system has flexible modeling capacity, which offers a principled means of encoding semantic, clinical expert annotations of colonic polyp tissue identification and segmentation. The learning generality to unseen data is bounded by boosting [12, 11] and stacked generality [14]. Extensive experimental results on polyp segmentation performance evaluation and robustness testing with disturbances (using both training data and unseen data) are provided to validate our presented approach. The reliability of polyp segmentation and measurement has been largely increased to 98:2% (ie. errors les 3 mm), compared with other state of art work [4, 15] of about 75% ~ 80%.

In this study we introduce a novel algorithm for learning a polyhedron to describe the target class. The proposed approach takes advantage of the limited subclass information made available for the negative samples and jointly optimizes multiple hyperplane classifiers each of which is designed to classify positive samples from a subclass of the negative samples. The flat faces of the polyhedron provides robustness whereas multiple faces contributes to the flexibility required to deal with complex datasets. Apart from improving the prediction accuracy of the system, the proposed polyhedral classifier also provides run-time speedups as a by-product when executed in a cascaded framework in real-time. We evaluate the performance of the proposed technique on a real-world Colon dataset both in terms of prediction accuracy and online execution speed.

In this paper we describe the architecture of a vision-based workspace monitoring system for different kinds of nonautonomous service robots. Based on a hardware system with a movable color CCD-camera, mounted on a linear sledge above the workspace, we are developing a hierarchical, modular software system with a flexible control which is not restricted to a single application. During the initialization a 3D map of the workspace is generated by tracking prominent features like edges and corners while the camera is moved. In the following, regions of interest are determined by means of motion detection and/or color classification. These regions are used for the search of known objects, e.g. humans, the robot itself, etc. After an object is identified, its posture and location is estimated by matching a geometric 3D model of the object with the detected regions. The feasibility of this approach will be proved with two applications.

Address block localization on mail pieces is an important task in postal automation. Unlike personal letters, business letters or postcards which have a high degree of global spatial structure among a limited number of entities, mail pieces of magazines and advertising do not have a preferred position of the address block which can be pasted in an arbitrary position and orientation. Items packaged in clear plastic, allowing the underlying print to show through, may cause further problems, as well as graphics, pictures and other printed text on such covers which also make the address block localization more complicated. In this paper, we present an approach for automatically locating the address block on covers of magazines, advertising letters and newspapers based on the recognition of the structure of a German address. First, we use a thresholding technique to decompose the image into smaller blocks. Then, these blocks are classified into characters and large characters. Afterwards blocks are searched which might be a corner of an address block. Starting from these blocks adjacent blocks are combined to connected regions line by line. These regions are candidates for an address block and are assessed in the last step. Heuristic rules and typical features are used to identify the correct destination address block among these candidates. Experimental results using images of different magazines, newspapers and advertising letters are given. Performance and speed are also indicated.

We present a fast approach for the detection of address blocks on images of oversized flat mail-pieces with a size up to 35/spl times/25 cm. In contrast to standard size envelopes the position of the address block cannot be restricted to the lower right region of the mail-piece. Furthermore the envelopes are mixed up with advertising, the return address, logos and other noise making the detection more complicated. In addition, for usage in a practical mail sorting system the recognition of the address block has to be accomplished within 200 ms for handwritten and machine printed addresses. In this approach a method based on the detection of inhomogeneous blocks is described. The image of the envelope is divided into small rectangular blocks for which a measure of homogeneity is evaluated. After determining an adaptive threshold inhomogeneous blocks are detected in a hierarchical search process. Adjacent inhomogeneous blocks are then merged to areas of arbitrary shapes. These areas are represented by the smallest enclosing rectangle. Several features are calculated for each area and a plausibility test is performed. The best scored area is assumed to contain the address block and is passed to an optical character recognition system. This approach meets the above mentioned time constraint on an i-860 based computer and has an excellent reliability.