Feature points repeatability on facial deformation

Feature point detection has been an important subject in image processing researches. It holds extensive potential applications in computer recognition, medical image analysis, artificial intelligence, and other fields. This paper explores and compares the performances of Harris corner detection and Scale Invariant Feature Transform (SIFT) methods in feature point detection works on facial image pre-processed using two different techniques; normal pre-process and background subtraction. Each method is performed to test their relation with repeatability. Three experimental stages have been executed; starting with feature point detection, identifying repeatability and finally measured the repeatability. The feature point is experimented on facial image surfaces with natural and smile expression. Based from the experimental result, we found background subtraction pre-processing technique give a good impact to the performances of both Harris Corner and SIFT detector in terms of searching the repeatability points.