论文部分内容阅读
Ｒａｎｇｅ图象的相关结构可被认为是受噪声干扰的分段光滑面．一旦曲面的参数被确定，那么就可以利用这些参数重建曲面，从而大大减少图象中的噪声．在Ｒａｎｇｅ图象中，几乎所有象素的统计特性往往与其邻近象素的统计特性相关．文中利用ＭａｒｋｏｖＲａｎｄｏｍＦｉｅｌｄ（ＭＲＦ）理论来模拟这种相关性，将曲面参数确定问题转化为一个后验均值求解问题．
The related structure of the Range image can be considered as a piecewise smooth surface that is disturbed by noise. Once the parameters of the surface are determined, the parameters can be used to reconstruct the surface, greatly reducing the noise in the image. In Range images, the statistical properties of almost all pixels are often related to the statistical properties of their neighboring pixels. In this paper, the MarkovRandomField (MRF) theory is used to simulate this correlation, and the problem of determining the surface parameters is transformed into a posterior mean solution.