Research On Model Matching Algorithms Based On Trusted Feature Distance Transform

Currently, the touch-screen devices have been widely used, but with the emergence of countless images on the Internet, if the image can be directly painted want to check out on the touch screen to retrieve the drawing board will be a direct exciting function, which is called sketch-based retrieval, the multimedia retrieval technology since the early 1990 s on the concern of researchers and scholars. Sketch retrieval now been applied to various fields, including the prevention of crime, digital libraries, photo sharing sites, search engines, medical diagnostics, GIS, remote sensing systems. It can help people of any age, can help children to understand the world, you can easily find help boys and girls T-shirt and floral skirt with a specific ornamentation, can help designers find the ideal image material.A novel image matching algorithm based on trusted feature in sketch-based retrieval is proposed. On the basis of traditional distance transformation(DT), trusted feature distance is introduced to eliminate those invalid distance values falsely contributed to the similarity calculation between sketch and database images, i.e. non-feature pixels which are actually far away from feature pixels will be ignored in similarity computing. Moreover, image similarity is weighted by matching feature density of the original color image to weaken the interference of a few highly matched pixels on the image similarity. Extensive experiments on various retrieval tasks show better accuracy than traditional DT methods. For a variety of complex network image, sketch retrieval purposes can be attributed to exclude unnecessary elements(small-scale features), leaving only the main components of perception and shape change in a nutshell is to enhance effective feature points and weaken the non-feature points.In order to evaluate retrieval effectiveness of the proposed method and to test the effects of parameter variations on the retrieval performance, a database of model and query images, i.e. the Caltech-101 data set, was used and several experiments were conducted. Experiment contains a total of eight types of sketch, through statistical sketch retrieval results to determine the optimal parameters, and then optimal parameters in thesame database after all four different algorithms, each algorithm were calculated ANMRR value by table it is clear that the search results based on credible sketch characteristic distance matching model algorithm was proposed in this paper based on the retrieval results than the traditional distance transform more outstanding.