基于显著性检测的双目测距系统

为了提高双目视觉测距系统中图像匹配的实时性与测距的精度，提出一种将显著性检测与焦距拟合相结合的双目测距方法。首先对双目相机进行畸变矫正，并利用双目相机成像的特点拟合相机焦距与目标距离的关系，随后对所得图像进行显著性检测，并提取目标区域，最后，利用surf算子对提取出的区域进行特征匹配，将匹配点代入测距模型中得到目标物体的距离。结果表明:显著性检测方法明显提升算法执行速度，焦距拟合降低双目测距模型误差，明显提升双目测距精度。

Binocular distance measurement system based on saliency detection

To improve real-time image matching and ranging accuracy in binocular vision ranging systems, this paper proposes a binocular distance measurement method, in which saliency detection and focal length fitting are combined. The method first corrects the distortion of the binocular camera and then fits the relationship between the camera focal length and object distance by using the characteristics of the binocular camera imaging and further applies the obtained image for saliency detection, and then the target object area is extracted. Finally, the surf operator is used to perform feature matching on the extracted region, and the matching point is substituted into the ranging model to obtain the distance of the target object. The results show that the saliency detection method can significantly improve the execution speed of the algorithm, and the focal length fitting can reduce the error of binocular ranging model, which significantly improves the binocular ranging system accuracy.