医学图象的识别与分析能够为临床提供定量比的诊断依据，而图象分割是其中最关键的一步。为提高医学图象侵分割效果，提出了一种基于特征距离的阈值分割算法，并将其与颜色特征分类相结合，来对眼科裂隙灯生物显微镜图象上的角膜充血区进行分割，分割结果可用于角膜充血区的定量体分析，另外，该算法中的样本典型值是通过一种三维直方图分块算法来确定的，实验结果表明，该算法可以有效地分割出角膜充血，其分割效果优于欧氏距离阈值法，且分析数据的精度能够达到临床诊断的要求。

Medical image recognition and analysis can provide quantitative evidence for medical diagnosis, while medical image segmentation is the first and significant step in this procedure. A novel feature distance based thresholding associated with color feature classification for ophthalmologic image segmentation is presented, and the segmented results can be used for the analysis of blood vessel region distribution on these images. Feature distance based thresholding can be widely used for segmenting the objects with any kinds of feature space structure, especially those with uncompact support, blurred boundary and uneven quantity, while Euclidean distance based method is only suitable for the sphere shaped feature space. Ophthalmologic image segmentation does not satisfy the conditions of Euclidean distance based method. In addition, a blocking algorithm based on 3D histogram to determine typical samples is proposed in this paper. Lots of experiments show that the algorithm in this paper can be effectively used in ophthalmologic image segmentation and good segmentation results have got. And the analysis results from segmented images meet the requirements of medical diagnosis.