An anisotropic diffusion model based on a log-normal distribution of a local gray-level is used to propose a way to denoise the panchromatic images. The implication of the low-edge gradient of the feature space for denoising and smoothing the noisy image is adaptively adjusted by the adaptive threshold parameter in a diffusion coefficient function. Furthermore, to terminate the diffusion process, an entropy-based stopping criterion is implemented. The proposed model is compared with the existing models such as Perona–Malik (PM), adaptive PM, difference eigenvalue PM, modified PM, and Maiseli–Gao. In order to analyze the performance of the models, quantitative metrics such as standard deviation, entropy, and the signal-to-noise ratio of a two-dimensional line profile are used. For further analysis, the results of denoising models are segmented using entropy-based segmentation techniques such as Harvda, Renyi, Kapur, and Yen models. A misclassification error metric is used to evaluate the segmentation results. The metric results show that the proposed model effectively removes the noise and preserves the features of a panchromatic image.