Research Of Corn Growth Stage Recognition Based On Computer Vision Technology

The progress of science and technology has prompted the development of agriculture to usher in the modern agricultural technological revolution.Information,new materials,biological and sustainable development technology will be integrated into one for agricultural intelligence,opened up the new way of agricultural production.As an important part of information technology,computer vision technology has been widely used in precision agricultural automation,especially in the applications of crop growth state detection and weeds identification.Computer vision technology has characteristics of non-destructive,high detection precision,high efficiency,flexibility and good real-time,avoiding the strong subjective and low efficiency weaknesses of the artificial observation.At present,there are some research results on the monitoring of crop development period,but the detection accuracy is not high and the universality is poor.Therefore,In order to overcome the above shortcomings,this paper uses the computer vision technology to study the maize seedling stage,the three-leaf stage,the seven-leaf stage and the tassel stage automatic observation algorithm,recording the date of corn arriving at each period,it is important to guide maize field farming activities.The main difficulty of automatic observation of maize seedling stage is that maize seedling is difficult to identify,and there are many interferences in maize field,such as soil and straw.Therefore,this paper first uses the image enhancement algorithm based on the retina model to enhance the maize seedling stage image,improving the contrast of the maize seedling,and then uses the improved super-green index to convert the enhanced image to gray image.After this,this paper uses the iterative threshold method to divide the corn seedling,and finally uses the relative change rate of the average number of seedlings per day to determine whether the corn into the seedling period.The contrast of maize seedlings and image background in maize three-leaf stage is low,not easy to distinguish.Therefore,a maize seedling enhancement algorithm is proposed to reconstruct the G channel value by using the RGB channel values of the maize trefoil image,and the new RGB image is obtained.In new image,the differences of corn seedlings and background are increased,easily to distinguish.And then the gray image of the new RGB image is obtained by processed by Fisher linear discriminant.Finally,the two-dimensional Otsu algorithm is used to divide the maize seedling.The method can well segment the corn seedling from the image under different conditions.The seven-leaf stage image segmentation algorithm of maize is the same as that of maize three-leaf stage.This paper chooses to use the color characteristics to identify maize tassel in corn tassel stage,but there are some differences in the color characteristics of maize tassel in different regions.Therefore,this paper proposes an automatic observation algorithm for maize tassel stage.The algorithm firstly enhances the color of tassel.The Cb and Cr component graphs are enhanced in the YCbCr color space,and then the YCbCr color is transformed into the RGB color space and tassels of maize are enhanced.The background color is different to tassels,and the tassels of different varieties are similar.After the enhancement,the improved Kmeans is used to cluster the grayscale images of the tassels of maize,and the area of the connected domains is used to remove the interference.The number of the maize tassels is larger than the set threshold indicating that corn tassel stage is arrived.Henan,Hebei,Shandong and Inner Mongolia are the main areas of maize growth in China.In this paper,the different maize development stages images of the four places are selected to verify the validity of the automatic observation algorithm proposed in this paper.The results are compared with the results of artificial observation.The results show that the proposed algorithm is feasible and effective.