區(qū)域分割的亞像素相位立體匹配算法
發(fā)布時(shí)間：2019-07-30 17:45
【摘要】：針對(duì)雙目光柵投影測(cè)量中深度不連續(xù)及邊界陰影處的誤匹配導(dǎo)致測(cè)量誤差大的問題,提出區(qū)域分割的亞像素相位立體匹配算法。該算法采用由粗到精的匹配策略,在標(biāo)準(zhǔn)極線幾何、相位等約束下獲得粗匹配點(diǎn)。根據(jù)變形柵線圖、展開相位圖的特點(diǎn)將粗匹配結(jié)果分為不包含邊界陰影的深度連續(xù)區(qū)域、邊界陰影及深度不連續(xù)區(qū)域,分別使用加窗匹配測(cè)度最小二乘法和外插法獲得亞像素匹配點(diǎn)。實(shí)驗(yàn)結(jié)果表明:該算法能有效消除傳統(tǒng)匹配算法在深度不連續(xù)處的視差平滑過渡;與傳統(tǒng)點(diǎn)基元匹配算法相比,該算法在邊界陰影及深度不連續(xù)處的平均高度相對(duì)誤差減小了62%~69%,與傳統(tǒng)模板匹配算法相比,該算法在邊界陰影及深度不連續(xù)處的平均高度相對(duì)誤差減小了57%~61%,在深度連續(xù)處的匹配精度提高了3~4倍。該算法有效提高了相位立體匹配的精度,對(duì)提高目前三維測(cè)量技術(shù)精度具有重要意義。
[Abstract]:A sub-pixel phase three-dimensional matching algorithm based on region segmentation is proposed for the problem of the large measurement error caused by the mismatch of the depth discontinuity and the boundary shadow in the binocular grating projection measurement. The algorithm uses a rough-to-fine matching strategy to obtain a rough matching point under the constraints of standard line geometry, phase, and the like. According to the deformation grid line graph, the rough matching result is divided into a depth continuous region, a boundary shadow and a depth discontinuity region, which do not contain the boundary shadow, and the sub-pixel matching points are obtained by using the windowing matching measure least square method and the extrapolation method, respectively. The experimental results show that the algorithm can effectively eliminate the disparity smooth transition of the traditional matching algorithm in the depth discontinuity. Compared with the traditional point primitive matching algorithm, the relative error of the algorithm at the boundary shadow and the depth discontinuity is reduced by 62-69%, Compared with the traditional template matching algorithm, the relative error of the algorithm in the boundary shadow and the depth discontinuity is reduced by 57% to 61%, and the matching accuracy at the depth is increased by 3-4 times. The method effectively improves the precision of the phase three-dimensional matching, and is of great significance for improving the accuracy of the current three-dimensional measurement technology.
【作者單位】： 西安交通大學(xué)機(jī)械工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金資助項(xiàng)目(11072183)
【分類號(hào)】：TP391.41
[Abstract]:A sub-pixel phase three-dimensional matching algorithm based on region segmentation is proposed for the problem of the large measurement error caused by the mismatch of the depth discontinuity and the boundary shadow in the binocular grating projection measurement. The algorithm uses a rough-to-fine matching strategy to obtain a rough matching point under the constraints of standard line geometry, phase, and the like. According to the deformation grid line graph, the rough matching result is divided into a depth continuous region, a boundary shadow and a depth discontinuity region, which do not contain the boundary shadow, and the sub-pixel matching points are obtained by using the windowing matching measure least square method and the extrapolation method, respectively. The experimental results show that the algorithm can effectively eliminate the disparity smooth transition of the traditional matching algorithm in the depth discontinuity. Compared with the traditional point primitive matching algorithm, the relative error of the algorithm at the boundary shadow and the depth discontinuity is reduced by 62-69%, Compared with the traditional template matching algorithm, the relative error of the algorithm in the boundary shadow and the depth discontinuity is reduced by 57% to 61%, and the matching accuracy at the depth is increased by 3-4 times. The method effectively improves the precision of the phase three-dimensional matching, and is of great significance for improving the accuracy of the current three-dimensional measurement technology.
【作者單位】： 西安交通大學(xué)機(jī)械工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金資助項(xiàng)目(11072183)
【分類號(hào)】：TP391.41
【相似文獻(xiàn)】
相關(guān)期刊論文 前10條
1 寧曉斐;胡波;趙磊;_澄鬧，
本文編號(hào)：2521040
本文鏈接：http://sikaile.net/kejilunwen/ruanjiangongchenglunwen/2521040.html
最近更新
教材專著
