Defogging of particle images is a common task in machine vision-based measurement of particle size distribution (PSD) for granular materials or products. Current defogging methods are challenged by particle images captured in industrial environment, where artificial light is dominant in the imaging process and the fog is often unevenly distributed in the image due to its rapid and random movement. The recovered images may contain serious overexposures and result in failures in PSD measurement. To solve this problem, we propose a novel model-based defogging method for particle images with different fog distributions. First, a physical model is established to describe the image formation in industrial foggy environment with artificial light. Then, an adaptive method is proposed to estimate the range of the key model parameter (light intensity coefficient) and to obtain candidate defogged images. Finally, the fog-free image is generated by fusing the candidate images to reduce overexposures while retaining the image brightness. The proposed method was validated experimentally using iron ore particle images and compared with ten current defogging methods. Results demonstrate that our proposed method is adaptive to different illuminations and different fog contributions. It outperforms other defogging methods in removing unevenly distributed fog from the image and thus greatly improves the measuring accuracy of particle size. The proposed image defogging method can be integrated into machine vision systems for particle-handling processes to facilitate PSD measurement in foggy environment. The sample implementation of the proposed method is available at “ <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/zhoushuyi/particle-image-defog</uri> .”