关键词: 图像去雾, 细节增强, 分层处理, 边缘保持

Abstract: Images of outdoor scenes show poor visibility and discernibility in the inclement weather conditions such as haze, smog and mist. This paper describes a new method for single image de-hazing which can enhance the details effectively. Based on the dark channel prior, it first blurs the fogged images with edge-preserving filtering, and then extracts the detail layer without fog. In the de-hazing process of the foggy image, the detail layer is appended to the preliminary de-hazing image produced by the median filter to get more meticulous and sharper images. In order to extract the detail layer of foggy image, it chooses the domain transform recursive edge-preserving filter, which can perform high-quality edge-preserving filtering of foggy images in real time.

Key words: image de-haze, detail enhancement, separated processing, edge-preserving