针对超声图象存在一种特殊的斑点噪声，使图象边界与细节变得模糊，而严重影响图象质量的问题，提出了一种自适应线边界检测方法，即以不同方向与不同大小的“窄条”来近似组织边界，这种“窄条”是指具有不同长度、方向各异的线段.该方法首先将经过每一像素点的“窄条”的最大投影作为该点的灰度值 ;然后应用假设试验优化方法确定“窄条”方向，而“窄条”的长度由基于区域增长的局部统计特性决定.通过引入噪声抑制与边缘保留量化参数，进行了仿真斑点图与组织超声图的处理实验，并与自适应斑点抑制滤波、直接线边界检测方法进行了比较，结果证明，该方法不但能有效地抑制斑点噪声、而且能很好地保留与增强图中的组织边界与局部细节.

There is a kind of noise called speckle in ultrasonic images. It degrades severely the quality of image, especially edges and details. An adaptive line boundary detection technique is presented. A stick is a line segment of variable length and different orientation. Sticks are short enough that they can locally approximate the edges and certain features in the image. In particular, tissue boundaries have the appearance of straight or gently curving line segments because they are cross-sectional views of the surfaces between tissue layers. The boundaries and certain details are smooth on the scale of spacing between scan lines and generally define closed convex shapes. The algorithm works by plotting at every point the maximum local projection onto the stick. The orientation of stick is decided by hypothesis test optimizing method and the length of stick is obtained by local statistics based on region growing technique. For quantitative evaluation and comparison with adaptive speckle suppression filter (ASSF) and line edge detector(LED), two parameters as measure of the noise suppression and edge preservation are introduced. It is proved that this method can not only suppress speckle noise but also preserve even enhance issue boundary effectively by applying it to the phantom and tissue images.