Face Recognition using a greedy based warping algorithm

摘要

Recognition of faces is an important and challenging problem of a biometric system. In this paper, a greedy based warping algorithm is proposed for face recognition, which is a variant of conventional dynamic time warping (DTW) algorithm. The algorithm is designed to improve the recognition accuracy and to reduce the time complexity of DTW. Many real-time image processing applications require a solution that is near-optimal. The proposed algorithm provides a near-optimal solution to face recognition applications with linear time and space complexity and gives a better recognition rate. In this, matching of test face image to a set of gallery face images is performed. The performance of the greedy based algorithm has been evaluated by the experimentation on standard face databases such as ORL and FEI. Experimental results show that the proposed algorithm performs well even if the face images have uncontrolled conditions such as expression, lighting, and pose variations.