视觉测量中环形编码标志点的精确识别算法研究

Research on Detection Algorithm of Circular Coded Reference Point in Vision Measurement

视觉测量中,在待测物体表面分布一些易于识别的标记,可以提高特征识别的准确性和精度,保证多幅图像间特征点对应匹配的可靠性。针对在投影角度或图像噪声较大时,环形编码标志点识别率低的问题,设计了一种环形编码标志点作为标记,并提出了一种编码标志点精确识别算法。该算法首先使用Otsu算法提取椭圆边缘环状领域,并对环状领域点进行椭圆拟合求解椭圆参数;然后根据圆度与椭圆参数约束排除图像中伪标志点;最后通过计算编码环段内椭圆的灰度梯度确定编码环带的编码值。所使用的圆心提取算法能达到亚像素级精度。试验结果表明,该编码标志点识别算法受投影角度、图像噪声等因素的影响小,可以实现编码标志点的精确定位与准确识别。

Using distributed coded points on the measured object is a reliable and common method for achieving the opti-mum target location and accurate matching of corresponding points among multi-view images. However, it is difficult to de-code the coded point while the image noise and the projection angle are large. Aiming at solve this problem, an accurate andstable detection algorithm is proposed. Firstly, an accurate ellipse detection which can get sub-pixel locations is adopted toextract targets, and several rules are used to remove the non-target points. Then, the gray scale gradient in the circular iscalculated to decode the target. Experiments show that the algorithm can identify and locate coded points accurately, and itis robust to the change of projection angles and noise.