Abstract: A kind of method with using of vertebra medical image to deal with image key data was presented. First, used of the method of piecewise linear transform to stretch gray level of image for enhancing contrast ratio between target area and background area or reducing the detail in the target area. Second, used of information of pixel gray level to set up weighing chart, choose target seed pixels and background seed pixels, and then make use of rand-walk algorithm to divide up image so that receive edge of contour line. At last, unit area pixel average gray-level in the range of edge contour line was computed. Through comparing gray value of unit area in the region with bone density which can be gotten from physic testing, the bone density may be received in the checked region.

Key words: Bone density, Piecewise linear, Rand-walk

摘要： 提出了一种对脊椎骨医学图像进行图像处理，分析关键数据求得骨密度大小的方法。该方法首先用分段线性变换对图像灰度拉伸，从而增强目标区和背景区对比度，减少目标区细节。其次，利用像素的灰度信息，建立加权图，选择目标种子点和背景种子点，用随机游走算法对图像分割。最后，在边缘轮廓线内部对照原图求得单位面积像素平均灰度值。该数据通过与大量的医学实验数据作对比可以得到准确的骨密度值。

关键词: 骨密度, 分段线性, 随机游走