二维几何图形测量中的边缘定位算法研究

Study on Edge Localization in Two-dimensional Geometric Measurement

研究图像优化测量和提取问题,针对CCD摄像机采集的图像中含有各种噪声,传统的边缘检测算法不仅不能很好地抑制噪声,而且定位精度只达到像素级。为了更好地提取图像边缘信息并抑制噪声,以提高测量系统的测量精度。在比较分析各种像素级边缘检测及亚像素细分算法的性能后,采用能同时协调图像边缘信息和抑制噪声的多结构多尺度形态边缘检测。根据测量系统的被测目标,采用矩估计亚像素细分算法来实现图像边缘的精确定位,并进行仿真。仿真结果表明算法通过提高定位精度,有效地提高了测量精度。

The traditional edge detection algorithm can not suppress noises,and the locating precision is at pixel level because of various noise contained in the CCD camera image.Many pixel and subpixel edge detection algorithms were compared and analyzed at this article.In order to obtain image edge information,inhibit noises and improve the accuracy of the measurement system,multi-scale and multi-structure mathematical morphology edge detection was used,because it can coordinate image edge information and reduce the noise.On this basis,according to the target of the measurement system,the moment estimation of subpixel subdivision algorithm was used to realize the precise positioning of image edge.The simulation results indicate that the algorithm can improve position accuracy,thus effectively improves the measurement accuracy.