Research And Implementation Of Cell Counting System Based On Image Processing

The number of fungus spore statistics,you can determine and analyze the disease of crop diseases,and early warning prevention and control,reduce losses and save investment costs.At present,domestic and foreign scholars to study wheat and rice diseases more,but the study of potato disease spore counting method is relatively small.Among them,the potato late blight due to wide distribution,epidemic fast,serious harm,has been a threat to China’s potato quality,high yield of one of the diseases.Due to the presence of potato late blight bacteria spore image of high adhesion,impurities and spores little gap and spore itself gray uneven characteristics,making the existing counting method for counting the results of its error is large,so the study of its counting method has important research value and significance.This thesis mainly focuses on potato late blight and uses image processing method to study spore image counting.Key research content and work include:(1)To build a cell counting system simple image acquisition hardware device,used to complete the online collection of spore samples.For the spores of the characteristics of the hardware of the main module is selected and determined,including industrial camera,microscope head,light source,etc.(2)The obtained spore microscopic image as the study object,the spore microscopic image segmentation process,extract the spores and adhesions spore separation.First,the image grayscale and filtering,and then according to the characteristics of the collected microscopic images of bacteria spores,using Otsu threshold segmentation method,iterative threshold segmentation method,K-means clustering method and adaptive threshold segmentation method to split the spore image experiments,the experimental results show that the adaptive threshold segmentation method segmentation effect is optimal.In this paper,the adaptive threshold algorithm is improved,and an adaptive threshold segmentation algorithm based on Integral graph is proposed to solve the problem of grayscale difference between Target and background due to uneven illumination and the segmentation error caused by the grayscale unevenness of the spore itself,and the speed of the algorithm is increased by about 60%.Then morphological treatment,through the comparison of the parameters of the experiment to get the optimal parameters;at the same time,the use of a combination of distance transformation and watershed algorithm,effectively inhibit over-segmentation and pseudo spore regions,to achieve the separation of adhesions spores.(3)The statistical number of spores isolated by using the connected area marker method,and the three cell counting methods were analyzed and compared to obtain the optimal counting method based on the adaptive threshold algorithm proposed by this topic,and the area,circumference and roundness of spores in the image were calculated,which provided data reference for agricultural workers to analyze disease severity.(4)Using MATLAB GUI compilation toolbox to complete the design of image processing based on cell counting system.The system software enables online and offline access to spore images,but also by clicking the button to complete the results of each processing flow of the spore image preview and automatic counting,the use of intuitive,simple operation,so that some non-professionals can complete the cell image professional processing.Finally,30 microscopic images of bacteria spores were used to test the cell counting system.Experimental results show that the average count accuracy of the system reached 99.08%.The average count error of the system is 0.92%,and the maximum count error is not more than 5%.