CAJ | 학술논문

针对室内场景双目立体匹配有别于一般场景立体匹配的特殊性,提出了一种计算简便、准确度高的立体图像匹配算法.该算法首先利用canny算子检测物体的边缘,根据边缘的线性不变矩寻找出目标物体,然后提取出目标物体轮廓的特征点,利用角度直方图计算出左右图像的旋转角度,最后利用角度向量实现左右图像的对应像素点的匹配.线性不变矩有效地将计算复杂度由二维降低到一维,大大降低了计算量.角度向量的提出降低了特征点匹配的复杂度,而且计算简便,准确率高.实验表明,该算法对图像的缩放、旋转、平移均免疫,具有较高的识别精度和良好的抗干扰性,计算效率高于传统方法,有着较高的应用价值.
침대실내장경쌍목입체필배유별우일반장경입체필배적특수성,제출료일충계산간편、준학도고적입체도상필배산법.해산법수선이용canny산자검측물체적변연,근거변연적선성불변구심조출목표물체,연후제취출목표물체륜곽적특정점,이용각도직방도계산출좌우도상적선전각도,최후이용각도향량실현좌우도상적대응상소점적필배.선성불변구유효지장계산복잡도유이유강저도일유,대대강저료계산량.각도향량적제출강저료특정점필배적복잡도,이차계산간편,준학솔고.실험표명,해산법대도상적축방、선전、평이균면역,구유교고적식별정도화량호적항간우성,계산효솔고우전통방법,유착교고적응용개치.
In view of the particularity of Indoor stereo matching, this paper presents an accurate and convenient stereo matching algorithm based on line moments and the angle vector. First, the algorithm finds the corresponding objects in stereo images by line moments, then calculates the rotation angle between matching contours by the angle histograms of the matching contours, and finally, finds corresponding matching points based on the angle vector. The complexity of calculation can be reduced from 2D to the 1D by using line moments, and the amount of calculation can be reduced greatly too. The angle vector is calculated simply and conveniently, making stereo matching accurate. It is proved by experiments that the stereo matching algorithm is immune from scale, rotation and translation.