Digital Media Image Enhancement Technology based on Data Analysis

In recent years, with the rapid development of computer and information technology, image processing technology is also developing continuously, and has become the object of mutual learning and discussion of computer science, medicine, biology, engineering, information science and other disciplines. In the image processing system, its processing process can be divided into three steps: the first is the original image preprocessing, the second is the image feature extraction, the third is the image identification analysis. Among them, the image preprocessing link, that is, the image enhancement. Therefore, before the analysis of the image, we must first improve the quality of the image, then we need to use image enhancement. Image enhancement is an important basic image processing method, its goal is to process an image, make it highlight some specific information, weaken or remove some unnecessary information, so as to obtain better visual effect and use, or transform it into a format more suitable for human or machine analysis and processing. Based on the above problems, this paper discusses how to construct the image super-resolution reconstruction model through the Convolutional Neural Networks (CNN) algorithm and the attention mechanism, and analyzes the performance of other algorithm models through experiments. The results show that the image reconstruction time based on CNN-AM is only 20.7 s, which is shorter than other algorithms.