In order to realize the detection and feedback control of the processing position of drip holes in drip irrigation tape with labyrinth emitters inlaid, the machine vision was used to detect the position of drip holes, and a optimized recognition algorithm based on multiple depth extraction of ROI from ROI in the image whitening area was proposed. Firstly, ROI was extracted preliminarily by removing the background. Secondly, ROI of drip irrigation tape image in whitening region was extracted by applying algorithms of global fixed threshold segmentation, dilation, closed operation and region growing. Thirdly, the global threshold segmentation algorithm based on histogram was used to improve the image edges, and the Hough circle transform was used to extract the position of drip hole, and the Hough line transformation and additional constraint conditions were used to detect the position of labyrinth emitter front edge.Finally, the relative distance from the drip hole to the front edge of labyrinth emitter could be calculated. The experiment results show that with the optimized algorithms, the accuracy of drip holes identification is over 98%, and the distance measurement error is reduced from more than 10% before optimization to about 5%.