Traditional coaches need to rely on experience and intuition to adjust athletes’ movements and postures, which has the problems of strong subjectivity and unstable effects. In this paper, an image optical processing method based on artificial intelligence is proposed to realize the adaptive adjustment of aerobics athletes. The research uses the image optical processing technology to collect and analyze the movement of athletes in the training process and extract the key movement parameters. Then, combined with artificial intelligence algorithm, the extracted action parameters are analyzed and compared, so as to achieve accurate evaluation and adjustment of athletes’ actions. Finally, the adjustment plan is applied to the training of athletes, and through real-time monitoring and feedback mechanisms, the athletes’ movements and postures are continuously adjusted and optimized. Through experiments and practical applications, the system can accurately determine the movements and postures of athletes, and provide corresponding adjustment plans based on the actual situation. Compared with traditional manual adjustments, the system has higher accuracy and stability, which can effectively improve the technical level and performance ability of athletes.