文献类型：期刊文献

中文题名：三维人工视觉图像特征提取仿真研究

英文题名：Three-Dimensional Artificial Visual Image Feature Extraction Simulation Research

作者：朱敏 苏博

第一作者：朱敏

机构：[1]贵州大学农业生物工程研究院;[2]广西科技师范学院科技处;[3]贵州理工学院信息工程学院

第一机构：贵州大学农业生物工程研究院,贵州贵阳550003

年份：2016

卷号：0

期号：7

起止页码：261-264

中文期刊名：计算机仿真

外文期刊名：Computer Simulation

收录：CSTPCD;;北大核心:【北大核心2014】；CSCD:【CSCD_E2015_2016】；

基金：贵州省科技厅自然科技基金(qian J word[2012]2012)

语种：中文

中文关键词：三维人工视觉;图像特征;多尺度海森法

外文关键词：3D artificial vision; Image feature; Mtdti-scale Hessian method

摘要：在人工视觉图像特征提取的研究中,为提高人工视觉图像的精度,利用两幅互相配准的三维测距图像和灰度图像,是图像特征提取的基础,而传统的随机过程理论算法只能单纯获取三维测距图像和灰度图像,没有相关的配准系数,导致在进行三维人工视觉图像特征提取时提取效果差的问题。提出一种三维人工视觉图像特征提取方法,通过多尺度海森法,引入Harris算子对三维人工视觉的图像局部特征进行提取。通过找到两幅图像相应局部特征点的仿射变换参数,使目标函数达到最小,通过判断是否满足阈值条件,实现特征点匹配。依据在两幅相应数字图像上互相匹配的像素点与标定好的摄像机参数,求出空间点在世界坐标系下的坐标,实现三维人工视觉图像特征提取。仿真结果表明,所提方法在三维人工视觉图像特征提取时具有较高的准确性和效率。
It can improve the precision of artificial visual image to take the advantage of 3D ranging image and gray image registered with each other during the research on feature extraction of the artificial vision image. In this paper, we proposed a feature extraction method for 3D artificial vision image. In the method, we introduced Harris operator to extract the local features of 3D artificial vision image thorough multi-scale Hessian method. Firstly, we minimized the objective function through finding the affine transformation parameters of the corresponding local feature points of the two images. Then we realized the feature points matching through judging whether the threshold condition is satisfied. Finally, we obtained the coordinates of the spatial points in the world coordinate system according to the pixel points matched with each other in the two corresponding digital images and the calibrated camera parameters to achieve the feature extraction of 3D artificial vision image. The simulation results show that the proposed method has high accuracy and efficiency during extracting feature of 3D artificial vision image.