Contrast Limited Adaptive Local Histogram Equalization Method for Poor Contrast Image Enhancement

Contrast limited adaptive histogram equalization (CLAHE) is a widely utilised method for image enhancement due to its speed and simplicity. However, this method faces two major limitations, namely, the requirement of manual parameter setting (i.e. clip limit and window size) and the use of a single fixed weight for enhancement, which lead to artifacts in some regions of the enhanced images. A number of approaches are used to overcome these limitations. However, each approach has its gaps, such as complexity, parameter tuning sensitivity, dependence on initial image quality and long computational time. Furthermore, most methods may perform poorly well on images with different types. This paper proposes a new modified version of CLAHE, called Contrast Limited Adaptive Local Histogram Equalization (CLALHE), to address its limitations. This method is designed to enhance image contrast locally and adaptively, without the need for user input. The CLALHE approach consists of several steps: First, multiple enhancements are applied to identify the optimal enhancement parameters. Second, these parameters are determined adaptively. Third, the original image is divided into subimages, and the optimal parameters are applied to each subimage independently, emphasizing and enhancing local features. Finally, the enhanced subimages are combined to produce the resulting image. The effects of local and adaptive concepts were systematically explored to identify the optimal parameters and validate the performance using three datasets (DIARETDB1, Pasadena-Houses 2000 and Faces 1999). Qualitative assessments demonstrated the excellent performance of CLALHE and showcased enhanced images with improved contrast, better-defined details and less time consumption. Quantitative evaluation further confirmed the efficacy of CLALHE, which surpasses other methods in terms of peak signal-to-noise ratio, entropy, absolute mean brightness error, structure similarity index, contrast improvement index and root mean square error, across various image types.