Eye Contour Extraction on Face Sketch Using Gradient Vector Flow Snake

Nowadays, modern life use biometric technology in face recognition such as to assist law enforcement and security. Automatic retrieval of photos of suspects from the police mugshot database can help the police narrow down potential suspects quickly. However, in most cases, the photo image of a suspect is not available. Therefore, a sketch drawing based on the recollection of an eyewitness is an alternative substitute for this matter. The process of facial recognition involves automated methods to determine identity, using facial features extraction such as eye contour, as an essential element of distinction. Two methods have been used in this experiment. First, find eye position on face sketch using template matching algorithm. Then extract the eye contour using GVF Snake. Template matching algorithm is used to find eye position or eye location point with or without grid. Instead of using grid, detection location of eye point position results 10 sketches of 50 sketches, 11 of 50 on large sketches, 8 sketches of 2° counterclockwise and 5 sketches of 2° clockwise unrecognized respectively. By using grid, there are 5 of 50 sketches, 2 errors on large sketches, 4 sketches of 2° counterclockwise and only 1 sketch of 2° clockwise unrecognized respectively. Furthermore, ROC test performance of template matching process, with 0.01333 thresholds, results 71% without grid and 77% using grid respectively. Finally, the eye contour extracted by GVF snake resulting over 60% right eye and 40% left eye cannot be extracted correctly and smoothly