CAJ | 학술논문

高程亚像素级视差信息的提取是空中立体视觉成像的关键，传统的亚像素高程信息提取采用变分法的立体匹配与光流估计方法，产生遮挡、辐射差异和几何畸变，高程立体成像的边界接缝处理不理想。提出一种基于变分原理的高程图像亚像素级匹配方法，采用连续过松弛拟合的边界接缝平滑处理方法进行修正，像素视差函数分为整数级视差函数和亚像素级视差函数，实现航空高程拍摄图像的亚像素级视差分析和三维图像重建。仿真结果表明，采用该方法实现亚像素配准和图像恢复，精度较高，稳定性和鲁棒性较好，在高空探测、空地目标识别等领域具有很好的应用意义。
고정아상소급시차신식적제취시공중입체시각성상적관건，전통적아상소고정신식제취채용변분법적입체필배여광류고계방법，산생차당、복사차이화궤하기변，고정입체성상적변계접봉처리불이상。제출일충기우변분원리적고정도상아상소급필배방법，채용련속과송이의합적변계접봉평활처리방법진행수정，상소시차함수분위정수급시차함수화아상소급시차함수，실현항공고정박섭도상적아상소급시차분석화삼유도상중건。방진결과표명，채용해방법실현아상소배준화도상회복，정도교고，은정성화로봉성교호，재고공탐측、공지목표식별등영역구유흔호적응용의의。
The extraction of high altitude the sub-pixel disparity information is the key to aerial stereo imaging, the tradi-tional method takes the stereo matching and optical flow estimation method, and radiation difference and geometric distor-tion is occurred. High altitude stereo imaging boundary joint processing is not ideal. We present a novel method of high alti-tude image pixel level based on the variational principle, and the successive over-relaxation fitting boundary is modified joint smoothing method, pixel disparity function into integer disparity function and the sub-pixel disparity function, the re-alization of aviation high altitude photographed image sub-pixel reconstruction and 3D image reconstruction are obtained. The simulation results show that, using this method to achieve sub-pixel registration and image restoration, it has high pre-cision, good stability and robustness, and it is useful in field sounding, and space target recognition.