Camera Response Based Nighttime Image Enhancement Using Concurrent Reflectance

This article proposes a novel scheme that enhances low-light images locally to prevail their details and concurrently adapts camera response models (CRMs) to maintain the naturalness of the image. Low-light images shadow significant information within them, which demands to be unveiled through image enhancement techniques. However, the contemporary methods prove deficient of one or more factors that determine a well-enhanced image. These methods inevitably include intensity distortions or globally enhance the image, which causes loss of local information. The method proposed through this article initially decomposes the image according to retinex theory by utilizing a measured adaption model. A novel CRM is utilized to enhance the illumination of the image, while a reconstructed histogram equalization method enhances the reflectance locally. The method has been devised with complete awareness of its practical compatibility for hardware or software purposes. Extensive quantitative, qualitative, and time complexity analyses confirm the supremacy of the proposed method over various state-of-the-art methods.