基于一种动态颜色模型的模糊C-均值图像分割

Fuzzy C-means image segmentation in a new dynamic color space

不同颜色的可见光本质上是具有不同波长范围的电磁波。本文试探性地提出了一种动态颜色模型,它模拟了成像曝光时间内图像平面所接收到的电磁波的动态变化。离散化之后,彩色图像的颜色特征能够被表示成一个K维矢量,称为彩色图像的动态颜色空间表示。然后建立了模糊C-均值分割算法,分别在动态颜色空间和RGB空间分割彩色图像,实验结果表明动态颜色空间的分割结果优于RGB空间的分割,从而验证了动态颜色空间的性能。笔者相信本文所提出的动态颜色模型也能够被用于纹理分析或其它的图像处理领域。

A dynamic color model was presented based on the fact that visible lights with different color are essentially electromagnetic waves with different wavelength. The model could simulate the dynamic change of the electromagnetic wave received by the image plane during image - forming exposure. The dynamic model can be diseretized, and then we used a vector to represent the color feature of pixels in a K - dimensional space, which is named as dynamic color space. To verify the dynamic space performance, an integrated fuzzy C - means segmentation algorithm was established to partition color images in RGB space and in our dynamic color space respectively. Experiments showed that the segmentation results in our dynamic space were better than that in RGB space. We believe that the dynamic color space can be used to texture analysis or other image processing.