针对线结构光测量系统中如何准确快速地提取光条中心这一关键问题, 本文提出了一种基于自适应模板的新方法。根据线结构光图像梯度分布特性, 通过高斯卷积获得光条各点的Hessian矩阵。根据图像点的法线方向构造中心提取的卷积模板, 并能够根据光条形态变化, 自适应地改变模板权重分布。通过切线方向采样构建修正模板, 有效修正噪声等因素造成的误差。算法仅对感兴趣区域提取, 大幅减少运算量。实验表明, 该方法具有精度高、对结构形态具有较高的敏感性、实时性较好等特点, 实现了工业应用中线结构光光条中心的高精度快速提取。

Aiming at solving the problem that how a line structure light measurement system can quickly and accurately extract the center line of laser stripe,a new method on account of direction template was proposed. According to the distribution characteristics of line structured light image, the Hessian matrix of the stripe points based on Gauss convolution was firstly obtained.On the basis of the normal direction of image points, convolution template for center point extracting was constructed. The weight distribution of templates could change adaptively according to the morphological changes of light stripe. The correcting template is constructed by tangent direction sampling to correct the error caused by noise. The extraction of regions of interest greatly reduces the computational complexity. The experimental results show that the method has the advantages of high precision, high sensitivity to the structure shape and good real-time performance, and realizes the high-precision and rapid extraction of the center line in industrial applications.