Authors : Seng Cheng Yeoh, Haidi Ibrahim and Theam Foo Ng
Abstract: Images that have been captured in dark environments may suffer from insufficient or uneven lighting conditions. This conditions mostly will make the captured images to have low contrast. Therefore, to reduce this problem, contrast enhancement algorithms have been suggested to improve the image quality. In this study, two contrast enhancement pipelines have been observed. Because histogram equalization method is an image enhancement method commonly used nowadays, these pipelines are using histogram equalization-based methods. These contrast enhancement pipelines have been developed by cascading Exposure Sub-image Histogram Equalization (ESIHE) with Contrast Limited Adaptive Histogram Equalization (CLAHE) in two possible sequences. ESIHE is one of global histogram equalization methods. It uses only one single transformation function, created from global statistics of the image. On the other hand, CLAHE is one of the local histogram equalization methods. CLAHE uses local transformation functions which are created from local statistics of the image. CLAHE applies histogram clipping to avoid extreme enhancement of the contrast. These pipelines have been evaluated based on four objective measures which are Noise Standard Deviation (NSD), Image Variance (IV), Speckle Index (SI) and Contrast Per Pixel (CPP). From the assessments by using 30 input images and comparisons with seven other histogram equalization-based methods, it is found that the contrast enhancement pipeline that utilizing ESIHE first than followed by CLAHE has performed better the other methods.
Seng Cheng Yeoh, Haidi Ibrahim and Theam Foo Ng, 2019. Contrast Enhancement Pipelines for Digital Image Captured in Dark Environment. Journal of Engineering and Applied Sciences, 14: 6145-6149.