【摘要】 图像分割和特征提取是图像处理学的经典课题之一。文中研究有效的手腕骨 Ｘ 光片图像特征提取方法，先对手腕骨 Ｘ 光片图像作预处理，利用灰度投影的方法对各手腕骨特征区域进行大致定位，获得区域性信息，缩小实际特征提取时的搜索范围；在各特征子区域中，基于形变轮廓提取边界的原理，引入初始骨模板的形状信息及特征子区域的边界条件作为骨轮廓提取的内部能量和外部能量判据对骨边界提取的迭代过程进行引导，有效地排除非骨边界信息的干扰，使处理结果符合预期的骨模板形状，从而达到提高图像分割速度和准确性的目的。更多还原

【Abstract】 An algorithm for segmentation of the finger bones in X\|ray images is presented. To locate the feature area coarsely, at first, those sub\|regions of image are isolated corresponding to the finger bones by using gray level projection technique, and then according to domain knowledge of the sub\|region, template matching techniques are used to initialize a deformable model to extract finger bone edge from the special sub\|region, the initial model is deformed subject to externally\|imposed constraints and its original bone shape infomation. By using the region\| and model\|based techniques, the segmentation speed and accuracy are improved. \;更多还原