Processing method for eliminating illumination unevenness of face image

Abstract

The invention discloses a processing method for eliminating illumination unevenness of a face image, and overcomes the defect that the illumination unevenness influences the face identification rate. The invention adopts the technical scheme that the processing method comprises the following steps: under the support of a computer and matched image processing software, adjusting an image threshold value by preprocessing gray values of pixel points, calculating dimension values of prescribed neighboring domain matrixes 3*3 and 5*5 of the pixel points of the image, overlapping and fusing the dimension values without weight, modifying and replacing the gray values of the pixel points in a white spot area by using centerline symmetry, finally carrying out mean-variance normalization calculation for a set of the modified image pixel points, converting the gray values of the pixel points into a gray level range of [0, 255] to form the face image, and displaying or outputting the face image. The method eliminates the influence of illumination change on face identification, and effectively improves the robustness of the illumination on the face identification so as to achieve high identification rate.

Description

Background technique [0001] The invention relates to a digital image preprocessing method, in particular to a processing method for eliminating uneven illumination of a human face image. Background technique [0002] At present, face recognition is a research hotspot in the field of image processing and pattern recognition, which involves pattern recognition, image processing, computer vision, machine learning and other disciplines. Face recognition and authentication technology has very important application prospects in national security, public security system and urban public safety. However, illumination changes reduce the face recognition rate, which is one of the bottlenecks restricting face recognition technology. How to effectively correct the illumination of the face image and reduce the influence of illumination is a crucial issue in applying face recognition to practice. [0003] Local Binary Pattern (LBP) is an image local texture descriptor without parameters...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the shortcomings and deficiencies of the prior art, the present invention proposes a preprocessing method for improving the uneven illumination of human face images. It is realized under the support, and the steps of realization include the image processing module and the LBP algorithm software operation support under the control of the CPU and the management of the main control program. The encoding calculation of the gray value of the pixels distributed at equal intervals is carried out to obtain the single-scale local binary operator set of the image, and then convert the output to form a face image. The method is implemented in the following steps:

Images