Comparative analysis of methods for assessing image contrast

The article provides a comparative analysis of image contrast estimation methods, including global methods (RMS-contrast, entropy contrast, range contrast) and local methods using convolutional operators (Sobel, Pruitt, Laplace, Roberts), as well as Weber contrast and local standard deviation. The features of each method, their sensitivity to noise, and the accuracy of estimating local and global brightness variations are investigated. Based on the analysis of experiments, it is found that local methods, such as convolutional operators and Weber contrast, are more accurate in estimating local contrasts but sensitive to noise artifacts. Global histogram-based methods, including RMS contrast and entropy contrast, provide an overall contrast estimate without considering local image features.