- 期刊
Gray-scale Image Colorization based on Conditional Deep Convolution Generation Adversarial Network
摘要
Current gray-scale image colorization methods generally have problems such as border blurring, loss of detail, and boring coloring. In response to the above problems, this paper proposes an improved conditional depth convolution to generate a gray-scale image colorization method against the network. The network model fusion generates a confrontation network structure, builds a discriminant network, introduces a deep aggregation structure network into the field of image colorization, and adds long connections on the basis of the traditional network, which improves the utilization of features while alleviating the problem of gradient disappearance, thereby improving The algorithm model's ability to process image boundaries and details, and dynamically evaluate the color quality of the image, alleviate the problems of boundary blurring, loss of details and boring coloring. Experimental results show that this method has certain advantages in improving the image quality of grayscale images after colorization.