Improving detection accuracy utilizing super-resolution

Abstract

In this paper, we propose a method for detecting a specific object that has been determined in advance. Further, we aim to improve recognition accuracy by utilizing a super-resolution method to generate a high-resolution image from a low-resolution image. To achieve these goals, our proposed method employs a total variation regularization filter, a shock filter, and a pulse enhancement filter. The experimental results show that our method generates accurate high resolution images, and improves the accuracy of distant object detection.