| With the rise of intelligent agriculture,precise pesticide application is gradually becoming a research hotspot of intelligent agriculture.Based on the conservation tillage plan of black soil in Northeast China,this paper studies the identification algorithm of seedling belt center line based on intelligent optimization algorithm and morphological algorithm for seedling belt strip tillage technology mode.The research of this algorithm aims to accurately locate the center line of seedling belt,so as to provide pesticide application decision for subsequent precise pesticide application.The main research contents are as follows:(1)Research on pre-processing algorithms for seedling strip images.Through reviewing related references,analyzing image grayscale and binarization algorithms,comparing the accuracy and performance between algorithms,and finally the image grayscale and binarization algorithms are determined in this paper.(2)Research on multi-threshold segmentation algorithm.According to Otsu single threshold segmentation principle,the multi-threshold segmentation principle of image is described,and different intelligent optimization algorithms are introduced to carry out multi-threshold segmentation.Through algorithm comparison experiment,the intelligent optimization algorithm with good performance is selected to achieve multithreshold image segmentation based on intelligent optimization algorithm.(3)Research and optimization of seedling center line recognition algorithm.The morphologic operation principle based on binary image is described,the binary image is optimized,and on the basis of the optimized image,pixel method,closing operation and other methods are used to normalize the seedling belt image.Then,the connected domain detection is used to find out the seedling belt region and identify the center line of the seedling belt.At the same time,the recognition algorithm is optimized for the special cases such as the adhesion of seedling and tape.(4)A GUI program based on the algorithm code of this paper was designed.The algorithm code of this paper was written using MATLAB software and the MATLAB GUI program was written based on the algorithm code.(5)Algorithm performance test.The algorithm identification and manual measurement were carried out on the center line of seedling belt under three kinds of height,and the algorithm identification results were compared with the manual measurement results.The comparison results showed that the qualified rate of the algorithm identification under 4m height is about 92.50%,and the average recognition time is 4.738 s,which is about 1/37 of the manual measurement time.At 8m height,the qualified rate of the algorithm is about 91.38%,and the average recognition time is5.585 s,which is about 1/57 of the manual measurement time.At 12 m height,the qualified rate of the algorithm is 91.25%,and the average recognition time is 5.776 s,which is about 1/85 of the manual measurement time. |
