Infrared weak target detection method and device integrating RPCA and deep learning

Abstract

This invention proposes an infrared weak target detection method and apparatus integrating RPCA and deep learning. The method includes: acquiring an infrared image containing at least one target; extracting the spatial location of the infrared image based on a configured robust principal component analysis algorithm to obtain coarse candidate regions for the target; performing semantic enhancement processing on the infrared image to obtain semantic information features of the target; and determining the location and category information of the target based on the coarse candidate regions and semantic information features using a pre-configured defined classification model. The method proposed in this invention, based on a target detection framework, constructs a framework structure that integrates a robust principal component analysis model and a deep learning model, effectively improving the target detection accuracy and fast, stable tracking capability.

Images