期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2017
卷号：5
期号：5
页码：9856
DOI：10.15680/IJIRCCE.2017.0505276
出版社：S&S Publications
摘要：Sketch based image retrieval systems takes sketches that are roughly drawn by the user as the input andretrieves similar image from a large data base of images. Ability to deal with sketches is the main advantage of thissystem over existing image retrieval systems. Similarity measurement between a sketch and an image is one of themain challenges of this system for which we propose a novel SBIR system which utilizes Saliency Matching. Thesystem uses the outlines in the image known as contour maps. Here we use the Salient contour map (SCM) for shapematching. SCM is utilized to identify salient objects inside the natural images with backgrounds and matches with thequery sketch. Along with SCM, we propose another feature known as Angular Radial Partitioning with Orientation(ARPO) feature. ARPO feature utilizes the edge pixels in SCM to extract the feature vectors. Sketch Based ImageRetrieval systems are very much helpful in the medical field, digital libraries, photo sharing library etc.
