Face Recognition Based On Nonlinear Cascade Gamma Transform

Purpose — At present,the face recognition technology in the field of computer vision is very accurate in an ideal environment,but in the actual application process,it is often susceptible to complex and uncontrollable external environment,resulting in a lower recognition rate.In order to reduce the impact of illumination on face recognition,this paper compares several common illumination normalization algorithms.In order to solve the shortcomings of traditional gamma correction,a face recognition algorithm based on nonlinear cascade gamma transform is proposed.This achieves the goal of eliminating illumination interference.Design/methodology/approach — Pre-processing the face image before face recognition,reducing the adverse effects of noise and shooting angle on face recognition,and then using the nonlinear cascade gamma transform algorithm proposed in this paper to respond to the pre-processed face image Transformation,and finally the processed image is passed through an improved PCA+LDA+SVM framework for face recognition.Findings — The related experiments prove that the improved PCA face recognition algorithm has a significantly improved speed compared with the traditional PCA algorithm under the premise of ensuring that the recognition rate is basically unchanged.Through the related experiments on the extended Yale B face database,it can be seen that using the nonlinear cascade gamma transform algorithm proposed in this paper to process the face image,the sensitivity of the image to illumination is reduced,and the accuracy of face recognition is improved.Limitation — The improved face recognition algorithm in this paper is only verified on the common face database,and these face libraries can only represent a part of real life,but in actual use,if more complicated lighting occurs,whether the algorithm can still maintain a high recognition rate,to be verified.Practical implications — After the face image is preprocessed by the nonlinear cascade gamma transform algorithm,the impurities on the original image are significantly reduced,and the contour of the face is more prominent,which makes the face recognition accuracy increase.Originality/value — In order to solve the shortcoming of the traditional gamma correction algorithm using the same gamma value for power law transformation in the global domain,a nonlinear cascade gamma transform algorithm is proposed.The algorithm uses a nonlinear transformation formula to assign a unique gamma value to each grayscale value,that is,the entire face image uses a varying gamma value for power-law operation to adjust the grayscale value of the pixel that improve focalization.