Defect Enhancement of Image Sensor Based on Multichannel Decomposition and Log-Intensity Entropy

In today’s smart world of the Internet of Everything, cameras are essential to the information acquisition system. The image sensor is a crucial component in camera production, and its quality directly impacts imaging quality. However, the vignetting phenomenon can interfere with defect detection in image sensors, leading to inaccurate results. To address this issue, this paper proposes for the first time a method using multi-channel decomposition and logarithmic intensity entropy to solve the vignetting interference problem. The proposed method first creates a vignetting model using Fourier functions to simulate changes in illumination. Then, by calculating the minimum logarithmic intensity entropy of target images, optimal parametric solutions that fit the variation laws of the vignetting function were obtained. Finally, multichannel decomposition was used to perform vignetting correction and defect enhancement of the image sensor. Experimental results demonstrate that this method effectively recovers luminance information from vignetting regions, corrects different degrees of vignetting phenomena, and enhances detail information about defects. This lays the foundation for subsequent defect detection of image sensors and will improve the accuracy of the overall system performance in the future.