Improving AdaBoost face detection quality and speed via cascade classifiers
Jinkui He, Ting Luo, Jinhong Jia, Shaoju Li, Yingxiang Liu, Zhaokui Yin · 2025
With the rapid development of intelligent human-computer interaction systems and people's increasing concern for safety, face-related detection and recognition technologies play an important role in the development of intelligent devices. Since face detection technology is a pre-technology for face recognition, the quality of the face detection algorithm determines the stability and reliability of the entire face positioning and recognition collected face image information is preprocessed, the image grayscale image is converted, and then the image histogram is equalized. The Sobel operator method is used to determine whether there are edge points in the image, and then the image is segmented; since image segmentation will increase the contrast between the target area and the background area, the maximum-minimum tracking method can effectively determine the boundary. Next, we select Haar-like features as face region features, extract features from preprocessed face images, and use the Adaboost algorithm to upgrade multiple weak classifiers to a strong classifier with high classification accuracy after introducing the integral graph. We then load the cascade classifier for testing to further improve the detection rate of the algorithm and reduce the detection time. This has important practical value for face detection. The face detection of this method can effectively guide the subsequent research of other algorithms such as face recognition.