Application of Intelligent Algorithms in Dynamic Fusion of Graphics and Video

Since videos contain dynamically changing lighting, texture, and depth information, image-video fusion often causes inconsistencies in color tone and light and shadow between images and videos. This study combines intelligent algorithms to solve the problem of inconsistency in multimodal data in dynamic image-video fusion. The study uses deep convolutional neural networks (DCNNs) to repair unnatural areas in image-video fusion and uses VDSR (Very Deep Super Resolution) to enhance the details of image-video fusion. Experimental results show that the SSIM of this method is 0.92, PSNR is 30.8dB, EPI is 0.85, and TC is 0.89, all of which are at excellent levels.