Photograph Based Pair-matching Recognition of Human Faces
Yao, M., Aoki, K., and Nagahashi, H. (2011). Photograph Based Pair-matching Recognition of Human Faces. Int. J. Comput. Electr. Autom. Control Inf. Eng., 5(12):1716-1723.
In this paper, a novel system recognition of human faces without using face different color photographs is proposed. It mainly in face detection, normalization and recognition. Foot method of combination of Haar-like face determined segmentation and region-based histogram stretchi (RHST) is proposed to achieve more accurate perf using Haar. Apart from an effective angle norm side-face (pose) normalization, which is almost a might be important and beneficial for the prepr introduced. Then histogram-based and photom normalization methods are investigated and ada retinex (ASR) is selected for its satisfactory illumin Finally, weighted multi-block local binary pattern with 3 distance measures is applied for pair-matching recognition. Experimental results show its advantageous perfo with PCA and multi-block LBP, based on a principle.