Image enhancement with naturalness preservation

Abstract

There exist many techniques in the literature to enhance poorly illuminated images. Most of these techniques are based on retinex theory and enhance images considering only the reflectance component of the image and ignore illumination component. However, illumination plays a significant role in achieving pleasing perceptual quality in an image. Also, naturalness of an image cannot be preserved without preserving its illumination. This paper proposes an image enhancement technique to enhance details of an image by keeping good illumination present in the image intact. For images with poor illumination, we propose to add suitable amount of brightness in low illuminated regions in the image to get pleasing perceptual quality and naturalness. There are two major contributions of this work. First, it proposes a bad illumination pass filter (BIPF) to locate badly illuminated areas in the image and subsequently create an enhanced image by highlighting details as well as good illumination areas. Second, it proposes an adaptive logarithm transformation to add the required amount of brightness in poorly illuminated areas in the image. Experimental results show that the proposed technique preserves the natural appearance of the image and produces better perceptual quality in comparison with existing state-of-the-art techniques. © 2018, Springer-Verlag GmbH Germany, part of Springer Nature.