在三维主动视觉系统中,为有效提取单幅彩色伪随机编码图像特征角点并给特征点匹配提供更多的样本信息,采用定位于彩色图像的Harris算子探测特征角点,并利用目标灰度同背景灰度比较构造一个自适应阈值判定条件,从而提高了角点响应的适应性和精确性。特征点提取实验及三维欧氏重构结果证实了该方法的适用性及有效性。
In the 3D active vision system, in order to effectively extract feature comer points of the single pseudorandom color coded image and to provide feature matching with more sample information, Harris operator orientating to color image is adopted to get an adaptive threshold to judge the comer point by comparing the gray of destination with that of the background. The results of the feature extraction experiment and the Euclidean 3D reconstruction confirm the applicability and effectiveness of the described method.