The volume of tree canopy provides theoretical basis for the orchard spray, and the application of airborne laser scanning is widely used in canopy volume measurement, but there is a problem of lack of canopy information. To solve this problem and improve the accuracy of tree canopy volume measurement, a method based on image processing to measure the tree upper and lower canopy volume ratio was proposed. A new M-K clustering method combining Mahalanobis distance and K-means algorithm was created to split the image target area and find the ratio of the volume of pixels in the upper and lower canopy by rotation integration method. The further research reduced the error (nearly 253%) measurement of unilateral canopy image processing on this basis. According to the estimation results of multiple images of different sides of the fruit tree by arithmetic mean method, M-K clustering method was modified, which became more accurate and stable. Totally 23 apple trees and 20 cherry trees were experimented in the orchard, and the results were compared with the artificial measurement results, which showed that the M-K clustering method was in good agreement with artificial measurement results with R2 apple of 0.775 and R2 cherry of 0.832. It can be used for the measurement of canopy volume ratio.