Abstract: Aiming to solve the problems of long period and low automation of granite moisture content detection and taking white granite as the research object, a moisture content detection method based on image features was proposed. The digital image processing technology was used to segment, normalize and denoise the white granite image to reduce the influence of different image brightness and poor color contrast. The correlation between the five image features of image gray histogram and water content was analyzed, and the linear regression analysis is carried out based on the least square method, The moisture content detection model of white granite based on image features was established. The model was verified by moisture content detection test. The results show that the mean value of detection relative error is 2.10%, the variance is 1.05, and the average detection time is 15.59 s, indicating that the detection model has high detection accuracy and fast detection speed.