Research On Detection System Of Rapeseed Pest Degree Based On Image Segmentation

Brassica napus is one of the important sources of seed oil.Also,it is one of the three oil rape varieties with the highest grain yield.It is widely planted in the middle and lower reaches of the Yangtze River,and is an important oil crop.However,during the rapeseed growth,diseases and insect pests will result in premature senescence of the plant,lower seed setting rate,leading to reduced production and oil.Therefore,the detection of disease and insect pests in rapeseed has become an important method to prevent crop yield reduction.However,manual detection of plant diseases and insect pests in rapeseed leaves and professional judgment requires a lot of preparatory work,and it is inefficient,time-consuming and labor-consuming,unable to achieve good real-time performance.In this paper,image processing technology is used to identify and segment the pest images of rapeseed and detect the degree of pest.The main research contents are as follows:(1)Commonly used several image segmentation methods,such as the characteristics of the threshold value or clustering and regional growth,edge detection and extraction for gray image segmentation results are good,but for the processing of color images are often effect is general,Grab Cut algorithm based on graph cut to use(color)in the image texture information and boundary(contrast),as long as the small amount of user interactions can get better segmentation effect.In view of the problem of poor segmentation effect of rapeseed leaves under the influence of natural light and other environmental factors,this paper adopts the Grab Cut algorithm combined with HSV space model image segmentation method to segment the target image of rapeseed leaves.The processed results will lower the error rate and achieve better image segmentation effect;(2)Using the connected component labeling algorithm to identify and mark the wormhole area in rape leaves,and then extracting the image features are my choices.The feature parameters included: leaf area,leaf perimeter,number of wormholes,wormhole perimeter,wormhole area,centrifugal distance,etc.Through the analysis of the characteristic values of all the leaf images,the characteristic parameters such as wormhole area and wormhole proportion were finally found through screening and combination,which could effectively identify the pest degree of rapeseed;(3)Design and implement a rape leaf pest degree detection system,the front end upload rape leaf image,through the background request server TXT file and analysis to obtain pest image related characteristic parameters,get rape leaf image pest degree results,visually display the rape leaf pest degree.The front end also shows the segmentation effect intuitively through the image of rape leaf after request processing.