SKETCH-FACE RECOGNITION USING HYBRID OF HOG AND GLCM FEATURE DESCRIPTORS

Abstract

Sketch-Photograph matching assumes an essential job in law enforcement to get offenders and a few years before sketch and photograph matching was done physically which was a tedious procedure. In past decade, different programmed sketch face matching procedures have been proposed to make the matching framework increasingly accurate and less tedious. Yet, issue with these frameworks are that in the event that they give great accuracy, at that point they exchange off it with high time utilization and computational expense and on the off chance that they attempt to lessen time utilization and computational cost, at that point they don't give great accuracy. So, to handle these issues an automatic system for sketch-face matching has been proposed in this thesis which gives great accuracy with lesser time utilization and computational expense. In our project, preprocessing is finished by changing over both the sketches and photographs to gray images and afterward processed with median filter to expel noise from the images while safeguarding edges in the images. The primary advance in our project includes feature calculation which has been finished utilizing hybrid of two different types of descriptors i.e. Histogram of oriented gradients (HOG) and Gray level Co-occurrence matrix (GLCM). After calculating the feature vectors for images, they have been utilized for matching purpose using Euclidean distance. Since, simple Euclidean distance has been utilized, time and computational utilization is extremely low in this project and hybrid of 2 distinctive feature descriptors gives great outcomes.