分割图像插值的一种局部算法

Local algorithm for segmented image interpolation.

对于医学序列切片 ,图像上像素之间的间隔常常小于切片之间的距离 ,而在图像处理中 ,常常需要它们有一致的分辨率 .对于已分割的切片图像 ,插入的切片图像也应是已分割的图像 .本文提出了一种图像插值的局部算法 ,该算法在产生新插入图像每点的分割信息与颜色信息时 ,计算只与相邻切片图像的局部像素值有关 .因此既保持了相邻切片的物体形状过渡自然 ,又具有良好的数值稳定性 .应用该算法于医学序列切片进行插值时 ,可达到良好的视觉效果 .

Usually in medical serial slices, the size of a pixel within a slice is much less than the spacing between the planes defined by the slices. But in image processing, they are often asked to have the same resolution in all three dimensions. Besides this, the interpolated images among segmented serial images should also be segmented. A novel local algorithm was used for the interpolation of serial segmented images. The method generated all segment and color information for each new interpolating point that is only dependent on its local blocks of pixels in its near slices. The shapes of the objects on the interpolated slices and their original adjacent slices changed progressively and naturally, and the algorithm was implemented reliably, experiments on medical serial slice image interpolation showed that the new algorithm can generate clear and accurate new pictures. This method was successively applied in a 3\|D Medical Image Visualization and Reconstruction System.