Modified Contrast Limited Adaptive Histogram Equalization for high dynamic range images

Tone mapping was originally used in the field of art where artists make use of a limited palette to depict high contrast scenes. The purpose of tone mapping is to display images or videos that correspond to the visual impression of an observer when watching the original scene by reproducing contrast and brightness between the original scene and the output device. However, the dramatically reduction of contrast and brightness induce a loss of contrast resulting in a loss of detail visibility. As a result, a fully automatic local tone mapping algorithm was introduced to increase the local contrast and reduce the loss of visual visibility. The algorithm was written by using MATLAB software. The local tone mapping method is based on the modified contrast limited adaptive histogram equalization (CLAHE) method. The tone mapping algorithm was first compressed using global tone mapping method. The image was then segmented into smaller parts using pixel-region-based image segmentation method. The segment regions were processed to become masks using morphological method (opening and closing method). Every mask’s clip limit was then decided according to the variance and mean. This step alleviates the artifacts introduced by the original CLAHE algorithm. Original CLAHE only have one clip limit and apply to each tile but the modified CLAHE method applied different clip limit in different mask in order to expand the dynamic range of the image. Lastly, the interpolation process was then applied to the entire image to remove the artificially induced tile border. For assessment, a subjective and an objective evaluation were conducted to evaluate the performance of the tone mapping algorithm. For subjective evaluation a survey was conducted to compare the fifteen images that had been processed by Reinhard’s,Drago’s, Ward’s and proposed tone mapping operator. For objective evaluation,visual difference predictor (VDP) was used to compare the difference between the HDR image and the tone mapped image. The generated probability map of the proposed tone mapped algorithm was then compared with the generated probability map of Ward tone mapping algorithm. Both results show that the proposed tone mapping algorithm produces images with good visual quality by retrieving more detail and local contrast in the image if compare with other tone mapping operator.

Preview PDF FK 2012 128RR.pdf Download (1MB) | Preview

Actions (login required)

View Item