A discrete particle swarm algorithm for geometric image inpainting

Reconstructing image structure can be accomplished using a geometric image inpainting algorithm. This is done following three main steps: the first step consists of locating the damaged region. In the second step, the structure of the corrupted contours is reconstructed by matching each damaged contour with its corresponding one in a way to obtain a visually plausible image. In the third step, the found couples are joined using a curve fitting technique. The search space of candidate solutions dramatically increases as the number of contours increases, which makes the search of the optimal solution by a traditional deterministic method infeasible. A good choice to solve this problem is the use of meta-heuristics. In this paper, a discrete particle swarm optimisation (DPSO) is used to find the best correspondence between contours using their curvature values as a quality metric. Additional information is used, which is the mean ordinate of each contour. It adds spatial information of the location of damaged contours. The experimental results and comparisons with the genetic algorithm show the efficiency of the DPSO for geometric image inpainting.

• discrete particle swarm optimization
• contour matching
• contour reconstruction
• curve fitting
• curvature
• metaheuristic optimization
• genetic algorithm comparison

• Metaheuristic Optimization Algorithms Research
• Generative Adversarial Networks and Image Synthesis
• Image Processing and 3D Reconstruction
• discrete particle swarm optimization
• contour matching
• contour reconstruction
• curve fitting
• curvature
• metaheuristic optimization
• genetic algorithm comparison
• Artificial Intelligence
• Computer Vision and Pattern Recognition
• Computer Science