Dynamic Visual Effect Optimization of New Media Art Under the Integration of Visual Perception and Deep Learning

Abstract

In today's digital and information age, new media art, as a new art form, is attracting more and more audiences with its unique charm. The integration of computer-aided design (CAD) and deep learning (DL) will surely breathe fresh life into the evolution of new media art. This article employs cutting-edge image processing technology and novel algorithm designs to enhance conventional dynamic visuals. We achieve this by incorporating image feature point detection to refine point cloud data registration, implementing a spatial bounding box to bolster matching point search efficiency, and refining image specifics and colour representation via categorical contrast adjustments and adaptive brightness corrections. Our findings reveal that, in contrast to traditional techniques, the optimized algorithm introduced here demonstrates superior efficiency and impact in object reconstruction and image enhancement, thereby notably elevating the visual appeal of new media artworks. The extensive applicability and practicability of the optimization algorithm are verified through comparative experiments and tests in different scenarios.