Aiming at the problem of low accuracy of personalized recommendation of popular science video in aerospace science and technology, a personalized recommendation method of popular science video resources in aerospace science and technology is proposed. This paper constructs a binary function combined with the user behavior data of popular science video to realize the user’s interest mining of popular science video of aerospace science and technology. It uses Ffmpeg to extract video frames, and uses the key frame algorithm to divide the categories of popular science video resources of aerospace science and technology. It applies the collaborative filtering algorithm and the optimization matrix decomposition model to design the personalized recommendation algorithm of popular science video resources. The experimental results show that this method has high accuracy of video classification, and can analyze the user’s interest with high precision, and improve the accuracy of personalized recommendation of science and technology popular science video resources.