Occluded Shape Matching for Image Database of Reduced Data Points

Occluded Shape Matching for Image Database of Reduced Data Points

  We present a new approach to typify object contours in a database by a reduced number of data points and to match object shapes in occluded conditions. For simplicity, contours are approximated by a set of quadratic Bezier curves and all control points are stored in the database. Initial contour segmentation procedure is carried out based on scale space parameters. The scale is chosen imposing the condition that a particular number of curvature zero-crossing points survive on the object contour for a longer period of its changes. Its value is neither too small nor too large so that the influence of noise effects and the loss of important features can be avoided. We introduce the distance matrix, which is constructed from curve to curve distance measurement between the test and database contours. The maximum alignment of distance matrix elements for their verification gives the total cost of similarity between the test and database contours. We present the experimental results to confirm the effectiveness of the proposed method and demonstrate the successful performance on object contours containing noise and occlusion.