摘要 : Images captured in backlit conditions (i.e., backlit images) often have a vast difference in lightness between bright and dark areas. In such a dark area in an image, the visibility becomes extremely low, making it indistinct to r... 展开 Images captured in backlit conditions (i.e., backlit images) often have a vast difference in lightness between bright and dark areas. In such a dark area in an image, the visibility becomes extremely low, making it indistinct to recognize the subject. Sufficient image quality cannot be obtained by simply applying a general image enhancement method to such a backlit image. Many methods specializing in improving the image quality of backlit images have been proposed to cope with this problem. Although these methods can effectively improve dark areas' visibility compared to general image enhancement methods, the enhancement process causes artifacts in bright areas. In this paper, we propose a single backlit image enhancement method that effectively improves only the visibility of dark areas while suppressing over-enhancement and artifacts. In the proposed method, the lightness of the output image is calculated by the weighted sum of the input lightness image and the enhanced lightness image based on a weight map. The enhanced lightness image is calculated by alpha-blending two lightness-converted images obtained by gamma conversion and histogram equalization of the input lightness image. The weight map is calculated based on edge-preserving smoothing with a guided filter of a binarized input lightness image obtained using Otsu's method. The experiment shows the proposed method's effectiveness by quantitatively and qualitatively comparing conventional image enhancement methods and the proposed method using various backlit images. 收起