The Solution Method Of The Sparse Representation And The Application In The Eyebrows Recognition

Biometric identification technology has been one of the hot spots in related field，it also is the key strategic technologies that countries are scrambling to develop.Compared with traditional identification methods such as cards， keys， accounts andpasswords， biometric technology is more convenient. At present， the biometrictechnologies used commonly include fingerprint， face， iris， palm， gait， ears，voice， handwriting and so on. Theoretically， any biological characteristics，including physical characteristics and behavioral characteristics， which areuniversality， diversity， stability and collectable， can be used for individualidentification.Automatic face recognition (AFR) has become one of the most active researchareas in computer vision and pattern recognition. In the long course of study， facerecognition has been a very good development， but in the theoretical andexperimental， light， corrosion， shelter conditions have a great impact on facerecognition algorithm. However， the eyebrows， which also have such features andperform an important role in the face， are rarely used as an independent biometric forrecognizing. Based on the pre-works， this paper does some further research in theeyebrow recognition. The main works are described as follows:1) We study sparse representation model. In the method of sparserepresentation， it needs to solve-min when solve sparse coefficient. Change theconventional continuous non-derivative function to the continuity and derivativefunction，when solve-min， it not only can solving the-min by derivationmethod，but also can approach0norm. First，how to define norm function whichhas to be differentiable， the second is how to derive norm minimization solutionprocess. On the basis of the definition of a function， then to solve the minimum of aobjective function which has constraints. Finally， against the problem of solving thecomplex nonlinear equations， to study the solving process combined with the theoryof optimization methods.2) We study eyebrow recognition based on sparse representation. The methoduses principal component analysis method to extract the eyebrows characteristics andrecognition eyebrows by the method of sparse representation. A number of experimentswere executed on the opened BJUT Eyebrow Database (BJUTED)， and this method gota high recognition rate which is up to about98.53%. Experiments on Color FERET FaceDatabase show that the eyebrow recognition rates are higher than that of face (when100 subjects， eyebrow recognition rate was98.8%， while face recognition rate was only96%). Those not only verify the feasibility of the eyebrow recognition， but alsopowerfully illustrate that the eyebrow recognition can replace face recognition in somecase.3) The comparison of sparse representation method and the nearest neighbormethod， nearest subspace methods and support vector machine method. Under thesame conditions， the sparse representation method， nearest neighbor method，nearest subspace methods and support vector machine method are applied to theeyebrows recognition， so as to arrive sparse representation method has betterrecognition rate and robustness.