Abstract: In view of the lack of high-dimensional features and inaccurate fusion and matching of panoramic video of complex railway passenger station, this paper proposed an image processing optimization technology of railway passenger station video fusion intelligent monitoring system based on deep learning. In this paper, scale invariant feature transformation algorithm was used to detect the key points of the image, convolution neural network was used to extract the high-dimensional features, random sampling consistency algorithm was used to eliminate the mismatch points, and the phantom problem was optimized to obtain better details effect. The proposed image processing optimization technology has been applied to Yangzhou East Station of Lianyungang-Zhenjiang high-speed railway. The application results show that this technology can effectively prevent image distortion and obtain better mosaic effect.