中文摘要：

为了准确识别和定位遮挡苹果果实,提出了基于圆曲率的苹果图像识别与定位方法。该方法首先对遮挡苹果图像用红绿色差法（R-G）分割,并提取苹果果实图像,然后对二值化图像进行形态学处理和边缘跟踪以获取目标轮廓,最后用最小二乘法对不完整轮廓进行圆拟合,提取出果实的圆心和半径。实验结果表明,该方法可以有效地对严重遮挡下的苹果进行识别和定位。

英文摘要：

Aimed to recognize and locate occluded apples accurately under different lighting conditions,a method based on curvature was presented.Firstly,the image was divided by color difference（R-G） to extract occluded fruit from original image.Secondly,image preprocessing algorithms such as Mathematical morphology and contour tracing were used to get the contours of apples.Lastly,least squares analysis was used for incomplete contours to fit circles which were used to locate the occluded apples.The experimental results show that this method was valid for heavy occluded apples.