Around the background of rapid development of intelligent technology, an efficient image processing system oriented to the smart management of human faces is focused on. The system is mainly developed towards high speed, high definition, high integration and reliability. The article investigates a novel automatic image processing algorithm, covering three key modules: automatic exposure control, color interpolation and chromaticity space conversion. The algorithm can process the output image of CMOS sensor in Bayer format in real time and adjust the image parameters to obtain a high-quality image. In terms of face recognition performance, the algorithm has a significant advantage in recognition speed compared with other algorithms, and the average recognition accuracy reaches 94.258%. In the practical application of ID card portrait processing, the image shows a more uniform grayscale distribution in the range of 5 to 255 after automatic color adjustment, and the color quality is significantly improved. Meanwhile, in the portrait enhancement experiments, the images obtained with this image automatic processing algorithm outperform the traditional ID card image processing methods regarding information entropy, mutual information, standard deviation and peak signal-to-noise ratio (PSNR).