Histogram Specification-Based Image Enhancement for Backlit Image
Yoshiaki Ueda, Daiki Moriyama, Takanori Koga, Noriaki Suetake
Abstract
In this study, focusing on the fact that the intensity histogram of a backlit image shows a characteristic bimodal distribution, we propose an image enhancement method for single backlit images using histogram specification. In general, well-known histogram equalization-based techniques are powerful in image enhancement, but when simply applied to backlit images, they often change the overall appearance and also highlight unnaturalness. In the proposed method, a target shape of the intensity histogram that improves the bimodal distribution is determined, and then intensity conversion by histogram specification is conducted while preserving hue and saturation. The effectiveness of the proposed method is verified through a series of comparative experiments.