Study On Technology Of Restore And Location Of Occluded Apples

Aimed to recognize and locate occluded apple fruit which harvested by robot,wespecially focused on the image of occluded apples and studied on recognization and locationof occluded apple fruit. Main contents and conclusions are as follows:(1) Studying on the image of apple fruit and the feature of background, and analyzingand comparing the denoising technology,we used Gaussian filting to eliminate some of noisesin the image.We also analyzed the grayscale peak of red apples and leaves and feature ofcolor difference (R-G). The results show that the image was segmented to extract occludedfruit from original image by color difference (R-G).(2) To remove small area of the image we presented method by means of seting to zeroto reduce the complexity of the algorithm,and method filling holes.We selected CannyOperator to detect apple fruit contour by analyzing and comparing Roberts Operator,SobelOperator, Prewitt Operator and Canny Operator. Experimental results show that fruit contourwere acquired accurately. (3) Studying on the feature of the apple fruits, we achieved results that a contour wasrepresented by circle analytical equations because it was an approximate circle, and we couldextracted smooth arcs from contour of apple fruit image by calculating the curvature ofsampling points.(4) Analyzing three kinds of methods fitting curve, we used Least Squares Analysis forfitting circles. The results which after image recognition and location show that: the averageerror rate is6.16%, which meet the needs of the harvesting robot localization.(5) Using VC++6.0and OpenCV we realised the main system-related modulesincluding recognizition and location of apple image and we analyzed the data and displayedthe analyzing results by VC++6.0and MATLAB mixed programming which providedconveniences for the the harvesting robot.