Shape-hased Image Retrieval Using a Non-Segmetntion Method Direction Histogram of Similar Edges

In this paper, we purpose a new algorithm for shape-based image indexing and retrieval. This new scheme efficiently extracts the image feature vector without any need to segmentation algorithms and high computation. The feature considers correlation between adjacent edges. Thus it includes information of continuous edges and lines of images and estimates shape of objects in images. With implementation of algorithms on a prototype system and using some evaluation metrics, we show the indexes are invariant relative to translation, image size, color, illumination, and small viewing positions. Experimental results show that averages of precision and recall rates of this new indexing scheme compared to other similar methods are 1.02 to 6.9 and 1.22 to 10.59 times greater respectively.