Year: 2024
Type: conference-paper
Abstract: : The high-quality apparent modeling of objects is usually done manually, which relies heavily on the experience of professionals, so the output efficiency is low. In recent years, automatic apparent modeling by machine has become a hot issue. Due to the difficulty of data collection, self-reinforcement learning method can be used to improve image quality, but this method still has the disadvantages of too redundancy and low efficiency. Therefore, the deep residual network is used to adjust its structure and improve the activation function. Finally, through the experiments of many different materials, the image results are compared based on the average peak signal-to-noise ratio and structural similarity (PSNR / SSIM). The apparent modeling performance based on depth residual network self-reinforcement learning is better and the function convergence speed is faster.
Authors: Yao Wang, Pengfei Guo, Haitao Zhang, Dong Huang
Institution: Shantou Polytechnic
Language: English
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4399257596