Sketch Image Retrieval System Using Canny Edge Detection for the Multiple Resolution Images

Abstract

In this paper, edge detection for the sketch image retrieval system is proposed. An effective edge detection is required for the edge-based sketch image retrieval system, because the retrieval results depend on edge images. Canny edge detection is known as one of effective edge detections, but Canny edge detection is difficult to apply to various images since it is needed to optimize three parameters. In this method, edges are, therefore, detected with Canny edge detection from multiple-resolution images of database images, and desired images are retrieved by extracting features from detected edge images. Experimental results of applying this method to 6,500 images in Corel photo gallery showed a good performance of this method.