论文部分内容阅读
本文给出一种较精确的光滑边界探测方法。首先对经过分类的二值图像执行生长与压缩运算以消除高斯白噪声并抽取目标粗边界;然后在粗边界内,依据一种全向梯度算子和方向信息跟踪边界元素;对这些跟踪得到边界元用一种局部比例估计算子估计各个边界像元上目标所占的比例;结合比例值和边界的前进方向信息计算边界曲线与边界像元的交点;对这些交点执行几何精纠正后用绘图机结合曲线光滑方法输出目标的光滑曲线及其面积。最后通过对模拟数据和实际图像数据的试验分析,表明本文给出的方法比以像元为单位输出的边界有更好的视觉效果,并给出更精确的边界位置与目标面积。这对于自动成图技术与图像处理技术的结合,对于遥感图像数据的直接成图输出,对于要求分像元精度量算目标面积的应用领域,对于低分辨率数字图像的大比例尺成图都有一定意义。
In this paper, a more accurate smooth boundary detection method is given. First, perform growth and compression operations on the classified binary image to eliminate Gaussian white noise and extract the target coarse boundary. Then, within the coarse boundary, trace the boundary elements according to an omni-directional gradient operator and directional information. Obtain the boundary for these traces The element estimates the proportion of the target on each boundary pixel with a local proportion estimation operator. The intersection of the boundary curve and the boundary pixel is calculated by combining the ratio value and the forward direction information of the boundary. After performing the geometric rectification on these intersections, The smooth curve of the target and its area are output by the machine in combination with the curve smoothing method. Finally, the experimental analysis of the simulated data and the actual image data shows that the method presented in this paper has a better visual effect than the boundary output in pixel units and gives a more accurate boundary location and target area. This is a combination of automatic mapping technology and image processing technology. For the direct mapping output of remote sensing image data, for the application fields that require the resolution of the pixel precision target area, large scale mapping of low resolution digital images has A certain meaning.