| As a kind of tea chrysanthemum,the Hangzhou White Chrysanthemum has been favored by more and more people because of its effects of evacuating wind and heat,calming liver and eyesight,clearing away heat and detoxifying.So far,the harvest of Hangzhou White Chrysanthemum is mainly done manually.Due to some factors such as short flowering period,large amount of harvest and high labor cost,Hangzhou White Chrysanthemum can't be harvested on time.Therefore,the rapid automatic picking of Hangzhou White Chrysanthemum has become an urgent problem to be solved.According to the current research status of agricultural picking robots and combined with image processing technology,machine learning theory,mechanical system design and other knowledge,in this thesis,Hangzhou White Chrysanthemum picking system based on binocular stereo vision and robot is proposed.The main research contents are as follows:(1)In order to improve the accuracy of image segmentation,image preprocessing is carried out firstly.And then the result of several preprocessing methods is compared.Finally,the bilateral filter is used to remove the noise in the Hangzhou White Chrysanthemum.(2)The feature of each area of the image of Hangzhou White Chrysanthemum was studied.In the experiment,the color feature and texture feature of the image are extracted respectively,and the color feature are represented by feature vectors composed of RGB three-channel pixels in the image.Then,the four feature quantities based on roughness,contrast,directionality and linearity in Tamura texture features were compared and analyzed.In this process,the data showed that the contrast feature quantity between Hangzhou White Chrysanthemum and soil is quite different,so the texture feature based on the Tamura texture feature is used to represent the texture feature of the Hangzhou White Chrysanthemum image.(3)The selection of the Hangzhou White Chrysanthemum classifier was studied.In this paper,the naive Bayes Classifier,Support Vector machine classifier and least squares support vector machine classifier(LS-SVM)are compared and analyzed respectively.The LS-SVM classifier shows high classification accuracy and low time consumption.Therefore,the background removal process of Hangzhou White Chrysanthemum is carried out using the LS-SVM classifier.(4)The transformation between the binocular stereo vision system and the coordinate system is introduced,and the binocular camera is calibrated by MATLAB software to obtain the internal and external parameters and distortion coefficients of the left and the right cameras.(5)The spatial orientation of the binocular vision system was studied.Firstly,the principle of similar triangle ranging is introduced.Then the algorithm in the left and right cameras is used to obtain the centroid,and the images of the left and the right sides are matched by the principle of centroid matching.Finally,the two centroid points are used to obtain the depth information of Hangzhou White Chrysanthemum.(6)Combined with the actual growth of Hangzhou White Chrysanthemum,this thesis designed a robot picking system of Hangzhou White Chrysanthemum,which mainly combined of the basic structure and control system,and then we designed a simple and efficient end-effector according to the picking requirements.(7)Using the designed robot system to carry out the picking experiment of Hangzhou White Chrysanthemum under different illumination conditions.According to the experimental results,the proposed algorithm can be used in three light conditions: smooth,backlight and shade.Through the experiments the picking successful rate is above 80%,and the average time taken to pick each Hangzhou White Chrysanthemum is 12.4 s. |
PDF Full Text Request