【摘要】 本文试图利用图像分割技术,实现叶片自动分类。为了充分利用像素的色彩,分割算法在RGB颜色空间进行。颜色空间数据量巨大,直接进行聚类效率太低,因此,本文运用一种特殊的存储结构存储颜色空间数据,按颜色的密度特征对图像中的颜色进行排序和聚类,并根据待聚类色彩与已有聚类中心距离是否小于类内最大距离来决定归入已有的类或形成一个新的类。实验结果表明算法具有较好的分类效果。更多还原

【Abstract】 We tried to employ the skill of image segmentation to distinguish herbal leaves automatically.In order to make full use of the colors of pixels,we conducted the segmentation in RGB color space.But one problem with color space is that its large data makes it inefficient when segment it directly.Therefore,we used a special storage structure to store data of color space,ordered and clustered the color of the image by density of colors,and decided which groups it belonged to by comparing the distance between the colors that are to be clustered and the colors that have been clustered and the maximum distance of the group.If the distance between the colors that are to be clustered and the colors that have been clustered is smaller than the maximum distance of the group,then it belongs to the same group.If the distance is bigger,then it belongs to a new group.The result indicates that this method of calculation has better segmental effect than the traditional one.更多还原