Design And Implementation Of Wheat Seedling Counting System Based On Image Processing

Wheat is one of the important food crops in China.The high and stable yield of wheat plays an important role in ensuring food security in China.A number of studies have shown that planting density is too sparse or dense can seriously affect wheat yield.Therefore,timely and accurate knowledge of the number of wheat seedlings has a positive significance for guiding production.At present,most of the seedling work is still done by manual methods,which are highly dependent on human resources and inefficient.Based on the image processing technology,the paper established a set of methods for measuring the number of wheat seedlings by using a mobile phone to photograph the wheat counting system.It solves the problem of high labor cost and low efficiency when manually calculating the number of wheat.The paper designed a wheat seed counting system based on image processing,which mainly includes the method of segmentation of wheat images,the counting method of wheat in seedling stage and the realization of several seedling systems.The specific research contents are as follows:Firstly,in the aspect of image segmentation,based on the characteristics of super green features,a segmentation method suitable for wheat in seedling stage is proposed.The algorithm of super-green feature and threshold segmentation is used to segment and process wheat images.The morphological algorithm eliminates and fills the burr edges and holes caused by image segmentation.Finally,the interference region with smaller area is filtered through the connected region to realize the effective segmentation of wheat images,which lays a foundation for the later seedling algorithm.Secondly,in terms of seedling statistics,a set of methods for counting wheat plants was proposed based on the analysis of wheat image characteristics during seedling stage.Firstly,the segmentation of wheat roots and stems was marked by the connected region labeling algorithm to determine the number and width of connected regions in the image,so as to grasp the basic information of wheat plant number;then clustering algorithm was used to cluster the widths of all connected regions to determine The adhesion area is drawn;finally,the waveform diagram is drawn on the adhesion area,and the accurate counting of the wheat seedlings is realized by extracting the characteristic extreme points in the waveform diagram.The counting method was tested by using 90 samples with different densities.The experimental results show that the average accuracy of the counting method reaches 91.09%.Finally,in terms of system development,the counting system adopts the C/S architecture,and the Android platform acts as the client.The client takes photos of the wheat and uploads it to the server.The server processes the wheat image and returns the counting result to the client.It is displayed on the interface of the App,thereby realizing the wheat counting system.Pass the test of the function and performance of the system.The processing time for processing an image is about 5s on average,which basically meets the needs of use.