Gray level transform model of ball image and its application in citrus image correction

In the process of quality inspection and automatic classification of fruit based on machine vision, fruit image is distorted due to the three-dimensional shape of fruit. Aimed to reduce this image distortion, the images of 57 balls, which were divided into three groups with the diameters of 75, 60, and 40 mm, respectively, and every group consisted of 19 balls with different colors, were captured at 6 different positions. The influences of size, color and image capturing position of a ball on the image gray level distortion were studied. Image gray level correction model of ball at the center of camera view field, and image gray level correction model of ball not at the center of camera view field, were developed, and the correlation coefficients R² of the models were 0.796 and 0.842, respectively. Based on these models, the gray level correction algorithm of citrus image was set up using MATLAB software, and the experimental results proved that the gray level correction model and algorithm were valid and reliable.