基于颜色特征和聚类的马氏距离图像分割法

给出了一种基于颜色特征和聚类的复杂彩图中进行目标图像分割的马氏距离算法.该方法利用目标的颜色进行图像分割.通过对彩图中的物体进行采样和分类,经过对每个像素点进行马氏距离计算和最小值寻找,将图像内的所有像素点进行归类,对目标图像与背景图像进行二值化分割,并对分类后含噪声的目标图像进行自适应滤波.设计了达到以上目的的人机交互式可视化计算机图像处理程序,对在水稻田中试验点上拍摄的水稻照片进行了分析处理,分离出了复杂背景下的水稻植株图像.实验结果表明,该算法能较好地解决复杂彩图中目标图像的分割问题.

Mahalanobis distance algorithm to separate objective picture based on color characteristic and color clustering

A kind of Mahalanobis distance algorithm is presented to separate the objective elements from a complicated color picture,based on the color characteristic of elements and color clustering.The picture is analyzed on account of the colors of target elements.After the elements is sampled and classified in the picture,the Mahalanobis distance ofeach pixeliscalculated.The smallest Mahalanobis distance value of each pixel is recorded and used to classify the pixels into different groups so that the target elements is extracted from the background in a binary picture.Then the binary picture is filtrated for removing noises by adaptive-filtering.A computer program with GUI(Graphic User Interface) is designed based on the algorithm to dispose the pictures taken from paddy field,and succeeded in separating the plant from the complicated paddy background.Test results show that the algorithm could separate target elements from a complicated color picture effectively.