Abstract: In this paper,through the analysis of the automatic tea picking process,the CMOS industrial camera is used to collect the tea image,and the gray level of the tea image is processed in the RGB color model,the Gaussian filter and Butterworth filter are used to filter the tea image,and the filtered image is segmented in the segmentation model.Finally,the clustering algorithm is used to identify the tender leaves and leaves in the image old leaves.The experimental results show that the correct rate of recognition is 89.5% for the young leaves in the image,85.3% for the old leaves in the image,and 75.6% for the old leaves and the young leaves in the image at the same time.