Machine vision is extensively utilized in diverse fields such as target detection and dimension measurement due to its inherent characteristics of high speed, precision, safety, and reliability, facilitating enhanced operational efficiency and quality assurance. However, influenced by factors such as processing technology and assembly techniques, non-uniform brightness and optical distortion significantly impact visual measurement results. To solve this issue, we propose a camera distortion correction method based on brightness non-uniformity compensation. This method preprocesses vignetted images using an improved simulated annealing algorithm with image entropy, processes the phase-shift and checkerboard patterns captured by the camera to obtain the intrinsic and extrinsic parameters of the camera, and corrects the distorted images. The proposed method significantly improves camera calibration accuracy and image rectification effectiveness compared to traditional camera calibration methods through advanced image preprocessing techniques. The experimental results demonstrate that after brightness non-uniformity correction, the error can be reduced by more than 80%. Using geometric distortion as an evaluation criterion, our method achieves an accuracy improvement of over 25% compared to Zhang’s algorithm. It exhibits versatility across domains including camera calibration, target identification, and detection, thereby augmenting the precision and efficiency of visual assessments. By implementing our proposed approach, advancements can be made in the refinement of visual measurement techniques, bolstering the reliability and efficacy of data analysis in fields reliant on visual data processing.