1.mean shift 的迭代过程只做了一次（为了简单）

2.对图像中的某一点进行平移时，只考虑了距离它小于2*hr的点，而不是图像中所有的 点。

3.实验过程中只用到了gauss核，虽然提供了method的选择，但在进行值域加权时只用 到gauss核，如果需要请自己改正。

4.meanshiftseg中，对分成一类的像素，取的是它们的平均值作为属于该区域的像素的 值。

5. 为了简单彩色图只做了第一维。其它类似的。-The iterative process 1.mean shift only once (for simplicity) 2. image pan when considering only one point less than the distance between its 2* hr point, instead of the image in all points. 3. The experiment uses only a gauss core, while providing a method of choice, but only used during the gauss range nuclear weighted, if necessary, please correct yourself. 4.meanshiftseg, the class of a pixel is divided into, and their average value was taken as the value of the area belonging to the pixels. 5. For simplicity only the first color image dimension. Other similar.