CAJ | 학술논문

针对传统灰度匹配算法精度和实时性无法兼顾的不足，提出一种基于图像金字塔结构和归一化互相关相似度函数相结合的区域分层匹配算法，主要可控参数包括匹配窗口尺寸、匹配特征值范围、视差阈值范围、相似度阈值等，并基于双目立体视觉系统，通过双目视觉标定、等效标准极线结构的图像校正等方法，实现了钢板表面的深度信息提取，实验证明算法具有较高的精度和效率。
침대전통회도필배산법정도화실시성무법겸고적불족，제출일충기우도상금자탑결구화귀일화호상관상사도함수상결합적구역분층필배산법，주요가공삼수포괄필배창구척촌、필배특정치범위、시차역치범위、상사도역치등，병기우쌍목입체시각계통，통과쌍목시각표정、등효표준겁선결구적도상교정등방법，실현료강판표면적심도신식제취，실험증명산법구유교고적정도화효솔。
To satisfy the requirement of both precision and real-time of traditional region matching algorithms,a region-based hierarchical matching algorithm based on the mean image pyramid was proposed combining with the normalized cross correlation method. The algorithm has been controlled efficiently through the parameters such as matching window size,the smallest eigenvalue,and the disparity search space,similarity measure threshold,etc. The eztraction of steel plate surface depth information was realized by the ways of camera calibration of binocular stereo vision system and image correction based on equivalent standard epipolar structure. The ezperiment results showed that the algorithm yields better accuracy and higher efficiency.