In current visual communication, information transmission is unclear, images are distorted, and processing efficiency is low, so this study uses IT image processing technology to solve this problem. The article establishes a deep learning based image processing framework, collects and annotates a large amount of visual communication image data, and uses convolutional neural networks (CNN) for image feature extraction and classification. At the same time, image enhancement techniques are applied to improve image quality, and processing time is reduced by optimizing algorithms. The experimental results showed that the image clarity of the system was improved to 0.98, with an average processing time of 1.54 s. The accuracy rate reached 92% in multiple visual communication tasks, effectively improving the information transmission effect. Research has shown that IT image processing technology can significantly improve the performance of visual communication systems, providing theoretical and practical support for future related applications.