Photo-To-Sketch Matching Using Gabor Wavelet Transform

Shaimaa M. Osman, Gamal Selim, Gouda I. Salama

Abstract: Face photo-to-sketch synthesis has helpful applications on each digital entertainment and law enforcement. This paper presents an efficient algorithm formatting sketches with a face images. in order to get the transformed sketch (synthesized sketch) that is in the same modality with the original sketch we use Roberts detection technique, extract features from both the sketch and the transformed sketch (output from edge detection technique), and finally the matching step achieved by using three methods that will be illustrated in section 6. The experiments achieved maximum accuracy of 94.3 % and with an excellent execution time for the sketch synthesis process of a photo about 5.5 seconds on a computer with 2.13GHz CPU

Keywords: LDA Linear Discriminant Analysis, Gabor Wavelet Transforms, Feature Fusion, SVM Support Vector Machine, Combining Classifiers