In order to measure the morphological characteristics of crop non destructively, the stem angle of tomato plant was measured by using digital image processing. The image of tomato plant was acquired in greenhouse and the image processing algorithm was discussed based on the geometric correction algorithm. The linear and bilinear model was applied and the results showed that the correlation coefficient (R2) between measurement of original image and plant was 0.50 with absence of geometric correction; when the linear and bilinear model was applied, the R2 was 0.625 and 0.723 separately. The corrected image was processed following the color space transformation, image segmentation, binary image filtering, straight line fitting and the angle calculation. The saturation image was selected and segmented by Otsu method. A good result was gained by combination the median filtering and close-open morphological method. After image thinning, the stems of the plant were simulated by straight lines based on least squares method. The angle between main stem and branch was calculated. The experiment result showed that there was linear relationship between the calculated by image processing technology and manual measurement with R2 of 0.703. The manual and automatic measurement result also showed good correlation with R2 of 0.985. It provided a method for the automatic measurement and analysis of tomato growth.