Research On Spray Quality Evaluation Indicators Detection Based On Machine Vision

Spraying pesticide is the most important method of pest and disease control,restricted by the spray method,spray machine and its parameters,the effective availability of pesticide is very low.And the method to detect spray quality is in need.Presently,there are many spray quality evaluation indicators detection methods,such as diameter detection based on the optical instrument,deposition distribution detection based on spectral imaging technique,deposition quantity deposition based on electric sensor,and so on.However these method are expensive or heavy or could only detect one indicator,not suitable for detection in field.To solve these problem,a spray quality evaluation indicator detection system is set up,and the algorithm based on this system is developed.Finally the detection of droplet coverage,coverage density,diameter distribution are detected.The main research results of this paper are as following:(1)Proposal for spray quality indicators detection system based on machine vision is developed with the basis of analysis the spray quality evaluation indicators detection research condition around the whole world;(2)Select water sensitive paper as the droplets collector which is a kind of paper that easy for layout and collection,the image acquiring platform based on smart phone camera function is set up.The whole hardware system for spray quality evaluation indicators detection system is developed and calibrated;(3)The algorithms are programmed with HALCON software.Binaryzation for image with unstable lightness is conducted with dynamic threshold method.Overlapped droplets segmentation is conducted with circulatory opening method based on circularity.With the algorithm of Huber,the droplets contour is fitted into circle to calculate the diameter;An air pressure droplets generating system is set up,and the diameter is detected with high-speed photography system.Comparing the diameter detected with machine vision method and actual ones,the diameter correction function is calculated;(4)The experiments for detection spray quality evaluation indicator are conducted.The coverage percentage detection result is greater than the results detected with constant threshold method and partitioned threshold method as 7.59% and 5.49%.The average relative error between coverage density detected with method in this paper and counting manually was2.87%.The detection results are also compared with the DIJ droplets analyzer detection results with coverage percentage greater 0.52% than it.And the average relative error of diameter distribution detection with method in this paper comparing with laser droplets size analyzer is 9.64%.This paper focus on spray quality evaluation indicators detection based on machine vision lay foundation for intelligent spray parameters detection,spray method improvement and pesticide effective availability increase.