基于双高斯拟合的金属表面光条提取算法研究

Extraction Algorithm of Laser Stripe on Metallic Surface Based on Double-Gaussian Fitting

针对线激光测量系统对金属表面进行测量时强反射光影响光条提取的问题,提出了一种基于双高斯拟合的光条提取算法。首先对光条截面的灰度进行分析,发现了光条灰度的多峰分布规律;然后推导光的反射模型,对金属表面强反射光的产生原理和能量分布模型进行研究;之后根据分布模型建立双高斯拟合模型,设计光条提取算法,并用样例验证了算法的可行性;最后进行对比实验,分析了双高斯拟合法与传统光条提取算法的提取效果,并对结果进行了置信度评价。结果表明:双高斯拟合法可以有效抑制光条图像中强反射光的影响,准确提取光条中心;双高斯拟合法的置信度评价优于传统算法。

A light stripe extraction algorithm based on double-Gaussian fitting is proposed to solve the influence of high-reflective light on a light stripe extraction when the line laser measurement system measures the metal surface.First,the multipeak distribution rule of light stripe gray level is determined by analyzing the gray level of the light stripe section.Second,the reflection model of light is deduced,and the generation principle and energy distribution model of high-reflective light on metal surface are studied.Then,a double-Gaussian fitting model is established based on the distribution model,and a light strip extraction algorithm is designed.The effectiveness of the proposed algorithm is verified using examples.Finally,a comparative experiment is performed to analyze the extraction effect of the double-Gaussian fitting method and the traditional light stripe extraction algorithm and to evaluate and analyze the confidence of the results.The results show that the double-Gaussian fitting method can effectively suppress the influence of high-reflective light in the light stripe image,accurately extract the center of the light stripe,and its confidence evaluation performance is better than that of the traditional algorithm.