Image retrieval system using a decomposition of shape information

In this paper, we propose an image retrieval system using the decomposition of shape information. First, the system extracts the boundary of the object using Minimum Error Thresholding and some morphological image processing method. After that, to extract the local feature, the image is divided into smaller areas. The feature relies on the discrete wavelet transform and singular value decomposition of the shape information. Next step, we compare the differences between the feature of the query image and all the images in the database. Finally, based on the differences measures and threshold, the best possible candidate matches are given as result. The experimental results showed that the proposed system effectively searched for the relevant images with various projective distortions.