A face quality evaluation method based on svm

Abstract

The invention discloses a face quality evaluation method based on SVM, and relates to the technical fields of image processing and machine learning. First, for an input image, set the key point coordinates of the image, detect the size of the input image, adjust the image to a square, correct the key point coordinates of the face, and convert it into a grayscale image. Then extract the directional gradient histogram feature vector of the entire image, and convert the grayscale image into an LBP feature map of the equivalent mode. Take five key points as the center on the LBP feature map, take a rectangle each, and extract five LBP feature histogram feature vectors. Splice the eigenvector of the direction gradient histogram and the eigenvector of the LBP feature histogram into a eigenvector, and input it into the pre-trained SVM model to obtain the quality classification of the face image, and obtain the eigenvector of the current image and the super The distance of the plane; calculate and output the quality score of this face image according to the quality classification and distance. The invention has high accuracy and low computing power requirement.

Description

Technical field [0001] The present invention relates to the technical field of image processing and machine learning, specifically a face quality evaluation method based on SVM. Background [0002] Biometric-based identity technology has been widely used in scenarios such as video surveillance, security, and human-computer interaction. The scale of China's biometric technology market is growing very rapidly and accounts for a large share in the world. Among them, face recognition has the advantages of non-aggression, non-contact and easy operation, and it is easy to obtain and has a very broad application scenario. [0003] However, in practical applications, the recognition rate of faces is greatly affected by the quality of faces. The factors affected by the so-called face quality include: the angle of the face during the shooting process, the brightness of the environment, the clarity of the imaging, and whether there is a mask on the face. A higher quality face should face th...

AI Technical Summary

Problems solved by technology

This type of method has a certain degree of accuracy, but it is generally slow and requires high hardware requirements, which is difficult for embedded devices or real-time detection

Images