Image Inpainting of Multi-Spectral Image with Laser Lines Based on Generative Adversarial Network

Abstract

Abstract This paper presents a Generative Adversarial Network based on image in-painting, which can reconstruct the shape using a multi-spectral image with a laser line. One of the difficulties in multi-spectral photometric stereo is to extract the laser line, because the required illumination for multi-spectral photometric stereo, e.g. the red, green, and blue lights, may pollute the colour of the laser line. In this paper, we presents a method, which uses the Generative Adversarial Network based on image in-painting, to separate a multi-spectral image with a laser line into a clean laser image and an uncorrupted multi-spectral image without the laser line, to reconstruct the shape using a multi-spectral image with a laser line. To make the proposed method applicable to real-world objects, a rendered image dataset obtained using the rendering models in ShapeNet has been used for training the network, and the evaluation shows the superiority of the proposed approach over several previous methods, on both rendered images and real-world images.