Using digital image processing techniques to diagnosis N nutrition has become one of the key tech-nologies. Since the application of digital image technology requires early data, a method image preprocessing, imagesegmentation method, extraction of the target RGB image based on MATLAB were studied in the paper. At first, theoriginal image was denoised using median filter. Then the segmentation work were carried out. The shadow imagewas removed using Otsu thresholding method. A specific threshold was selected using HIS color model H-channelimage for soil segmentation. YCb Yr color model was used in Cb channel, a specific threshold selected in segmenta-tion of the whiteboard. Finally, the green leaves RGB image was gotten. R, G, B averages of all the pixels were ob-tained using MATLAB programming. This provided data to support the follow-up beet nutrition diagnosis and creat-ed a viable prerequisite.