基于图像处理的雾滴沉积分布试验研究

Experimental study of droplet distribution based on image processing

针对我国喷药作业以经验为主,过量使用农药的问题,采用超红算子2R-G-B与OTSU算法结合的图像处理方法,研究叶片表面形态特征、雾滴粒径和叶片倾角3因素对雾滴沉积分布的影响。结果表明:1)该图像处理方法能有效分割雾滴与叶片背景,基于该方法的雾滴覆盖率测定结果与人工分割方法比较,相对误差<5%。2)叶片表面形态特征、雾滴粒径和叶片倾角3因素对雾滴覆盖率的影响均显著。3)在相同喷量条件下,雾滴覆盖率随叶片倾角增大而减小,但当倾角增大到30°时,粒径为138.2和157.6μm的雾滴在粗糙和光滑叶片表面上的覆盖率发生突增。4)雾滴覆盖率随粒径增大而减小,但当叶片倾角大于60°时,粒径为157.6和197.1μm的雾滴在相同叶片表面上的覆盖率差异不大。5)在相同叶片倾角条件下,粒径较小的2种雾滴的覆盖率在粗糙叶片表面上最大,在光滑叶片表面上最小。所提出的基于图像处理的雾滴覆盖率测定方法,可为雾滴在叶片表面上的分布研究提供评估手段。

Aiming at the problem of the present pesticide spraying applications mainly depending on irrational experience and cause excessive use of pesticides, the objective of this study was to analyze the factors that affect the droplet distribution on plant leaves to provide scientific basis for spraying. The leaf surface structure, droplet sizes and the leaf orientation were investigated. An image processing method, which was the combination of the Excess red algorithm 2R-G-B and the OTSU algorithm, was proposed. The results showed that:1) The image processing method was able to efficiently extract the droplet area, and the relative error of the coverage rate calculation based on this method was less than 5% compared to the result of the manual segmentation method. 2) The significance of the three factors to droplets distribution was obvious. 3) Under the same spraying flow, droplet coverage area tended to decline with the increase in droplet sizes in general, and the coverage rate trend of two smaller droplets on rough and smooth surfaces had a turning point when the inclination angle was 30°. 4) With the increasing of leaf inclination angle, the droplet coverage rate generally decreased;However, when the surface inclination angle exceeded 60°, the coverage ratio of two bigger droplets did not show an obvious difference regarding the same leaf surface structure. 5) When the droplet diameter was 138.2 or 157.6 μ m, the droplet coverage ratio on the rough leaf surface is biggest while the ratio on the smooth leaf is smallest. In conclusion, The droplet coverage calculation method based on image processing proposed in this study could provide an evaluation approach for studying the droplet distribution on leaves.