Research On Strawberry Recognition Technology Based On Machine Vision

The economic value of strawberries is very high,but its ripeness is short and it is not easy to store.It needs timely picking and transportation.At present,the strawberry picking in China depends on artificiality which is inefficient.This restricts the development of the strawberry planting industry.Therefore,it is urgent to implement strawberry picking automation.The primary task is to make the machine automatically identify the strawberry position.In this paper,ridge strawberries were taken as the research objects,and the related research was carried out around strawberry image recognition.The method of strawberry recognition based on machine vision technology was proposed.The main research contents are as follows:(1)The intelligent detection technology of fruit recognition at home and abroad was analyzed,and various color models in machine vision were studied.The RGB and HIS color models were adopted in this study.The specific technical route of this study was put forward: the general camera was taken as the image acquisition hardware,the MATLAB software package was taken as the image processing software platform.and the strawberry recognition technology based on machine vision was developed.(2)Using image of strawberry model,the image preprocessing techniques such as denoising,segmentation,enhancement and feature extraction of strawberry images were studied.In the strawberry edge extraction technology,the edge detection technology based on Sobel operator was selected to meet the actual needs of the study by comparing the actual results.(3)A large number of real images of planting strawberries were collected in the field.The K-means clustering algorithm was used to achieve the segmentation of the fruit image of the strawberry image.Then the edge of the strawberry image was identified and fitted,and other morphological operations were performed to find the position of the strawberry.(4)On the image data of the real strawberry,the HOG characteristics of the strawberry image were extracted,positive and negative samples were created,and the SVM(Support Vector Machine)classifier was trained so that the strawberry imagecould be correctly judged,and then The NMS(non-maximal suppression)algorithm was used to traverse the entire image containing strawberries using the threshold window to correctly identify the strawberry position.