Optimalisasi Reduksi Noise Menggunakan Chain-Code Termodifikasi pada Pendeteksian Wajah

Abstract

Digital image face detection had been developing so much in this last decade. The searching for the best method is still walking until today. This face detection research is using color model based segmentation method combined with template matching. There are 4 main process in this research. First, motion detection is used to minimize the size of the image by subtracting image n with image n+m. Then, by using YCbCr color model based segmentation, the image is separated by skin color area and non-skin color area. Third, is to reduce the noise by using the blurring way and by filtering its wide. The last one is matching the area that last with template that has been prepared to detect which area is face. The result of the research, from 217 images and 10 videos , shows that this method reach up to 70,5% face detection accuration  procentage.