(1. 湖南科技大学 信息与电气工程学院，湖南 湘潭，411201；
2. 广东嘉铭智能科技有限公司，广东 广州，510600)
摘 要: 针对USB-C产品尺寸AOI(automated optical inspection，自动光学检测)生产线上3D激光检测系统中因图像不清晰造成针脚平面度误判率高的问题，从激光原图像入手探究其原因，结合三次样条插值多项式的光滑性，研究基于三次样条插值函数的3D激光图像去噪算法。该算法过程为：首先分析3D激光测量仪获取图像的过程，从获得的图像中分离并提取2D信息；然后，对图像检测区域逐个判断可疑噪声点，再针对每一行以位置为横坐标、像素值为纵坐标，采用三次样条插值函数计算噪声点的实际像素。研究结果表明：所提出的算法与分数阶积分算法相比，能够在保留边缘特性和纹理信息的同时，更好地去除噪声；该算法实用性强，能减少误判，大大提高工作效率。
关键字: 3D激光；去噪；三次样条插值；分数阶积分；平面度检测
（1. College of Information and Electrical Engineering, Hunan University of Science and Technology, Xiangtan 411201, China;
2. Guangdong Jiaming Intelligent Technology Co. Ltd., Guangzhou 510600, China）
Abstract:Aiming at the problem of high misjudgment rate of pin flatness caused by unclear image in 3D laser detection system of USB-C product size AOI(automated optical inspection) production line, according to the original laser image, a 3D laser image denoising algorithm based on three splines interpolation function was studied combined with the smoothness of cubic spline interpolation polynomial. Firstly, the process of acquiring image by 3D laser measuring instrument was analyzed, and then 2D information was separated and extracted from the acquired image. Then, the suspicious noise points were judged one by one in the image detection area. For each row, the position was the horizontal ardinate and the pixel value was the longitudinal coordinate. The cubic spline interpolation function was used to calculate the actual pixel value at the noise points. The results show that, compared with the recent fractional integration algorithm, the noise can be removed better by the proposed method which can retain edge characteristics and texture information.The proposed algorithm is practical, and it can reduce misjudgments and greatly improve production efficiency.
Key words: 3D laser; denoising; three splines interpolation; fractional integration; flatness detection
