Image Sampling Using Q-Learning

Abstract

: With the advent of the digital information age and multimedia technology development, the amount of image data is increasing day by day. The method of image sampling has been paid much attention to. The traditional triangular mesh sampling method needs to initialize the sampling set and the metric tensor before sampling, which is prone to problems such as unreasonable specification. Therefore, an intelligent image sampling method based on the Q-Learning reinforcement learning algorithm is proposed. Built on the interaction between reinforcement learning agents and the environment, an adaptive sampling method is designed to update agents' characteristics constantly. The experimental results show that this method can achieve the same effect as the traditional triangular mesh sampling method and is more intelligent.