Deep-sea Nodule Mineral Image Segmentation Algorithm Based on Pix2PixHD

• DOI: 10.32604/cmc.2022.027213
• Publication Date: 2022-05-18T06:24:40Z
• Authors: Wei Song; Haolin Wang; Xinping Zhang; Jianxin Xia; Tongmu Liu; Yuxi Shi

Deep-sea mineral image segmentation plays an important role in deep-sea mining and underwater resource monitoring evaluation. The application of artificial intelligence technology to projects can effectively improve the quality efficiency mining. existing deep learning-based algorithms have problems such as accuracy rate is not high enough running time slightly longer. In order performance images, this paper uses Pix2PixHD (Pixel Pixel High Definition) algorithm based on Conditional Generative Adversarial Network (CGAN) segment images. model a coarse-to-fine generator composed global generation network two local enhancement networks, multiple multi-scale discriminators with same structures but different input pictures generate high-quality test results datasets show that identify more target minerals under certain other conditions. evaluation index shows improves recall compared CGAN U-Net algorithm. It for expanding learning techniques field exploration