| China is one of countries facing serious water shortages.More and more areas are droughty because of global warming in recent years.So water conservation is increasingly urgent. As we know,agricultual water-using accounts for three-quarters of the total water consumption in China,so agricultual water-saving is very significant.Dripping irrigation pipe is one of the most common agricultual water-saving products.lt transports water to the roots of crops for irrigation by the dripping heads embeded in the pipes. These dripping heads needs to be punched holes. No holes or deflected holes in dripping irrigaiton pipe can cause the deaths of cropper locally. At present,manual work, which is low-efficiency and has high ratio of error, is used on dripping irrigation lines. So how to detect holes in a high-speed and more effective way is very important.Based on the theory of computer vision,a research on detection for dripping irrigation pipe holes is made in paper. Firstly, the paper introduces the conception and application value of computer vision.Then the functions and composing of the detection system are presented.The relative algorithms which involve on image pre-processing,image binarization,Hough transform and ROI extraction are emphatically discussed and analyzed in paper.Finally,the paper designs and realizes a method of holes detection after the study of image characteristic. Automatic extract of ROI1 and ROI2 which are two regions of interst in the image,and compute the region of holes ROI3 by detecting the edges of dripping trough.Then detect the location of the hole.The relative location between the edges and the hole is obtained. The method reduces the processing datas by using multi-ROI technique.And it improves the automatization level by the adaptive-threshold ways of Canny algorithm and Hough transform.The mass experiments shows good results and practicability of this method. |
