SKIN DETECTION USING FUZZY LOGIC

Abstract

With an increasing emphasis on simulating human intelligence into machine intelligence, the future demands advancement in the artificial intelligence technology. One of the major application areas in machine vision is detection of objects. Image processing mainly addresses the issues of detection and recognition of different prospects in an image. Due to random nature of images, it is not possible to covereach and everything with 100% accuracy but an effort can be made by adding some fuzziness so thatit works for almost everything. Thisresearch presents a study and implementation of a skin detection system based on combination of different color spaces using fuzzy logic. A simpler approach using boolean logic is also presented which is, later on, enhanced using fuzzy logic. In the past, YCbCr and HSV have been extensively used for skin detection but during our research the discrepancies (false alarms) showed the limitations of these basic skin detectors. In order to reduce false alarms , three color spaces YCbCr, HSV and ROB have been used to extract the general skin regions and the results of the three are combined using a fuzzy inference system. Results reveal an accuracy of 97.30% for skin detection and a significant reduction in therate of false alarms as compared to the boolean approach