Underground personnel face recognition method based on multi-scale unified LBP operator

Aiming at the problem that the traditional face recognition algorithm is vulnerable to illumination and dust in the identification of underground personnel, it is difficult to adapt to the complex environment of underground and affect the recognition accuracy. It is proposed to identify the face of underground personnel based on multi-scale partitioned LBP operator. The method retains the advantages of unified LBP operator, and extends the traditional LBP operator by using the concept of multi-scale partitioning. By extracting facial image features at different scales, the global and local grasp of facial images is enhanced and weakened. The effect of local noise on the overall recognition accuracy. Using the partition processing histogram to obtain the composite feature vector, the singularity of the traditional algorithm is compensated, and the multi-scale histogram feature vector is obtained. In the image classification, the similarity of the facial feature histogram is judged by the chi-square test. The threshold value compares the feature vector of the acquired image with the feature vector of the image to be matched stored in the system to realize face recognition of the underground personnel. The experimental results show that the LBP operator is unified by multi-scale partitioning, and the texture feature information of the facial image of the underground personnel after the histogram equalization is fully and effectively extracted, which can comprehensively reflect the facial features of the underground personnel; Compared with the human face recognition algorithm, the multi-scale partitioning unified LBP operator has significantly improved the robustness of noise and illumination. Combined with image partitioning, the multi-scale unified LBP operator has better grasp of the local and overall image. Good results, the proposed method in the face recognition of the underground personnel, the accuracy rate of 94.25%, compared with the traditional well below recognition algorithm increased by 5%, this method can improve the recognition accuracy in the identification of underground personnel.