关键词: 葡萄, 葡萄, 微喷弥雾调控, 产量

Abstract: To solve the problems of low accuracy of human counting and time consuming of photoelectric counting and so on in laboratory test of single soybean plant, the computer vision system and the MATLAB software development platform were used in single plant test instead of human counting in laboratory test of single soybean plant. The algorithm used the spatial filtering of the soybean grain image to remove the noise, and the "Otsu" method optimized the global threshold segmentation of the image. Based on the image processing, the two factors, the number of grain and the size of grain, were measured. Two indexes, soybean grain counting and soybean grain size grading, were studied with six soybean varieties as materials, namely ‘Zhouhei soybean’,‘Zhouqing soybean’,‘Zhoudou 11’,‘Zhoudou 18’,‘Zhoudou 22’and‘Zhoudou 23’. The experimental results showed that the algorithm and the procedure were accurate and effective, and the grain number of single soybean could be calculated accurately. The algorithm and procedure determined that the average area of the soybean grain in each variety was positively correlated with the 100 grain weight, and the coefficient of determination was 0.987. So the algorithm and procedure can determine the size of soybean grain accurately and effectively. In short, the method of soybean grain detection based on MATLAB image processing can reduce the human labor intensity and human visual deficiencies relatively, and it is of certain significance in improving work efficiency and accuracy.