【摘要】 针对汽车儿童安全座椅弧形槽内部结构复杂、几何尺寸不易测量的问题,提出一种基于SURF算法和CSS算法的弧形槽几何尺寸视觉检测算法。首先,双目相机倾斜拍摄弧形槽图像,检测弧形槽图像SURF特征并提取优质匹配集合实现图像配准,以渐入渐出法融合弧形槽图像。然后,采用基于改进CSS算法对弧形槽轮廓曲线进行角点检测,利用角点分割弧段进行弧形槽尺寸检测。实验结果表明,所提出的方法实现了弧形槽宽度和长度参数测量,宽度值测量误差<2%,长度值测量误差<1%,SURF图像拼接算法平均处理时间2.50 ms,改进CSS角点检测算法平均处理时间0.30 ms,满足弧形槽检测的实时性和精度要求。更多还原

【Abstract】 Because the internal structure of the arc-shaped groove in the child safety seat is complicated and it is not easy to measure its geometric size,a geometric size vision measurement algorithm based on SURF and CSS is proposed. The binocular camera is used to collect the images of the arc-shaped groove. In order to achieve the image registration,high-quality matching sets are filtered from the original features of the images and detected by SURF algorithm. The integral arc-shaped groove image is fused with the fading in and out algorithm. Then,its contour curves are segmented with corners and detected by improved CSS algorithm,and then,the geometric size is measured. The research results indicate that the proposed method is used to measure its width and length. The measurement error of its width is less than 2%,and the measurement error of its length is less than 1%. The average processing time of using the SURF image stitching algorithm is about 2.50 ms and the average processing time of using the improved CSS corner detection algorithm is about 0.30 ms. It meets the requirement of the real-time accuracy of the arc-shaped groove measurement.更多还原