For automatically navigating and identifying branches obstacle in the picking process of agricultural harvesting robots，it is necessary to solve the defection of iterative threshold segmentation since the gray scale difference between target and background is not clear. The iterative threshold segmentation of apple branch images based on contrast limited adaptive histogram equalization (CLAHE) was proposed. Firstly，the RGB color space of the apple branch images were transformed to the XYZ andI1I2I3 color space by transformation, and the X-Y color difference factors and I2 color factor of the apple branch images were extracted to analyze their gray level difference. Then the CLAHE was applied to the image in which the gray level difference was not obvious before iterative threshold. Finally, the apple branch images were segmented from the original images. Results showed that the ratio of successful segmentation was 92%.