基于最小错误率贝叶斯决策的苹果图像分割

为了实现苹果分级完全自动化，对苹果图像的分割进行了研究。依据最小错误率贝叶斯决策理论，提出了一种基于最小错误率贝叶斯决策的图像分割方法。从图像的直方图中估计出服从正态分布的不同类别参数，对图像中每一像素点进行不同类别判断。通过对多幅图像试验，取得良好的分割结果。试验结果表明，该方法无须滤波而具有良好的抑制噪声的能力，在图像分割中是一种可行的方法。

Apple image segmentation based on the minimum error Bayes decision

To accomplish the goal of completely automatic apple grading, the segmentation ways of apple image were analyzed. Based on the minimum error Bayes decision theory, the authors proposed a new way of image segmentation. Various parameters obeying normal distribution were estimated from the histogram and the pixels were judged to different sorts. Good segmentation results were obtained from several testing images. The results demonstrate that this method does not require any filter and has better ability in restraining interference. It is a feasible way for image segmentation.