Bundle adjustment is a camera and point refinement technique in a 3D scene reconstruction pipeline. The camera parameters and the 3D points are refined by minimizing the difference between computed projection and obse...Bundle adjustment is a camera and point refinement technique in a 3D scene reconstruction pipeline. The camera parameters and the 3D points are refined by minimizing the difference between computed projection and observed projection of the image points formulated as a non-linear least-square problem. Levenberg-Marquardt method is used to solve the non-linear least-square problem. Solving the non-linear least-square problem is computationally expensive, proportional to the number of cameras, points, and projections. In this paper, we implement the Bundle Adjustment (BA) algorithm and analyze techniques to improve algorithmic performance by reducing the mean square error. We investigate using an additional radial distortion camera parameter in the BA algorithm and demonstrate better convergence of the mean square error. We also demonstrate the use of explicitly computed analytical derivatives. In addition, we implement the BA algorithm on GPUs using the CUDA parallel programming model to reduce the computational time burden of the BA algorithm. CUDA Streams, atomic operations, and cuBLAS library in the CUDA programming model are proposed, implemented, and demonstrated to improve the performance of the BA algorithm. Our implementation has demonstrated better convergence of the BA algorithm and achieved a speedup of up to 16× on the use of the BA algorithm on various datasets.展开更多
为了有效抑制 H．264/AVC压缩视频流的传输错误扩散，提出一种基于率失真优化的 H．264参考帧选择算法。该算法针对 H ．264/AVC中的多种预测模式，在像素级准确估计了差错环境下的视频传输失真，并将该失真模型与率失真优化准则结合，...为了有效抑制 H．264/AVC压缩视频流的传输错误扩散，提出一种基于率失真优化的 H．264参考帧选择算法。该算法针对 H ．264/AVC中的多种预测模式，在像素级准确估计了差错环境下的视频传输失真，并将该失真模型与率失真优化准则结合，在率失真框架内选择最合适的参考帧，使解码段的失真度达到最小。实验结果表明，改进算法相比传统的预测编码方法PSNR值大约提高了1～2dB，能有效改善 H．264视频传输的抗差错性能。展开更多
AIM： To investigate cap morphology after small-incision lenticule extraction （SMILE） and its effects on intraocular scattering. METHODS： Sixty-five eyes of 33 patients undergoing SMILE were enrolled. In addition ...AIM： To investigate cap morphology after small-incision lenticule extraction （SMILE） and its effects on intraocular scattering. METHODS： Sixty-five eyes of 33 patients undergoing SMILE were enrolled. In addition to regular evaluation, Fourier-domain optical coherence tomography was used to investigate cap thickness at ld, lwk, 1 and 3mo postoperatively. The optical quality including modulation transfer function cutoff frequency, Strehl ratio, Optical Quality Analysis System （OQAS） values, and objective scattering index （OSI）, were evaluated using OQASTM. RESULTS： Cap thickness decreased from ld to lwk （P〈0.001）, but remained higher than intended thickness of 120 μm after 3mo （P〈0.001）. Cap thickness in central area was thinner than that of in the paracentral and peripheral areas （P〈0.0001）. Total number of microdistortions decreased from ld to 3mo （P〈0.0001）. Pearson analysis revealed a weak correlation between OSI and standard deviation of cap thickness at ld and lmo, as well as between range of cap thickness and OSI at 1mo. No correlation was found between microdistortion and OSI, but a negative correlation existed between microdistortion and range at ld and lmoafter surgery. CONCLUSION： The corneal cap tends to be more accurate and regular with time lapse. Better cap morphology tends to contribute less intraocular scattering in the eyes undergoing SMILE.展开更多
Rectification for airborne linear images is an indispensable preprocessing step. This paper presents in detail a two-step rectification algorithm. The first step is to establish the model of direct georeference positi...Rectification for airborne linear images is an indispensable preprocessing step. This paper presents in detail a two-step rectification algorithm. The first step is to establish the model of direct georeference position using the data provided by the Po- sitioning and Orientation System (POS) and obtain the mathematical relationships between the image points and ground reference points. The second step is to apply polynomial distortion model and Bilinear Interpolation to get the final precise rectified images. In this step, a reference image is required and some ground control points (GCPs) are selected. Experiments showed that the final rectified images are satisfactory, and that our two-step rectification algorithm is very effective.展开更多
文摘Bundle adjustment is a camera and point refinement technique in a 3D scene reconstruction pipeline. The camera parameters and the 3D points are refined by minimizing the difference between computed projection and observed projection of the image points formulated as a non-linear least-square problem. Levenberg-Marquardt method is used to solve the non-linear least-square problem. Solving the non-linear least-square problem is computationally expensive, proportional to the number of cameras, points, and projections. In this paper, we implement the Bundle Adjustment (BA) algorithm and analyze techniques to improve algorithmic performance by reducing the mean square error. We investigate using an additional radial distortion camera parameter in the BA algorithm and demonstrate better convergence of the mean square error. We also demonstrate the use of explicitly computed analytical derivatives. In addition, we implement the BA algorithm on GPUs using the CUDA parallel programming model to reduce the computational time burden of the BA algorithm. CUDA Streams, atomic operations, and cuBLAS library in the CUDA programming model are proposed, implemented, and demonstrated to improve the performance of the BA algorithm. Our implementation has demonstrated better convergence of the BA algorithm and achieved a speedup of up to 16× on the use of the BA algorithm on various datasets.
文摘AIM： To investigate cap morphology after small-incision lenticule extraction （SMILE） and its effects on intraocular scattering. METHODS： Sixty-five eyes of 33 patients undergoing SMILE were enrolled. In addition to regular evaluation, Fourier-domain optical coherence tomography was used to investigate cap thickness at ld, lwk, 1 and 3mo postoperatively. The optical quality including modulation transfer function cutoff frequency, Strehl ratio, Optical Quality Analysis System （OQAS） values, and objective scattering index （OSI）, were evaluated using OQASTM. RESULTS： Cap thickness decreased from ld to lwk （P〈0.001）, but remained higher than intended thickness of 120 μm after 3mo （P〈0.001）. Cap thickness in central area was thinner than that of in the paracentral and peripheral areas （P〈0.0001）. Total number of microdistortions decreased from ld to 3mo （P〈0.0001）. Pearson analysis revealed a weak correlation between OSI and standard deviation of cap thickness at ld and lmo, as well as between range of cap thickness and OSI at 1mo. No correlation was found between microdistortion and OSI, but a negative correlation existed between microdistortion and range at ld and lmoafter surgery. CONCLUSION： The corneal cap tends to be more accurate and regular with time lapse. Better cap morphology tends to contribute less intraocular scattering in the eyes undergoing SMILE.
基金Project (No. 02DZ15001) supported by Shanghai Science and Technology Development Funds, China
文摘Rectification for airborne linear images is an indispensable preprocessing step. This paper presents in detail a two-step rectification algorithm. The first step is to establish the model of direct georeference position using the data provided by the Po- sitioning and Orientation System (POS) and obtain the mathematical relationships between the image points and ground reference points. The second step is to apply polynomial distortion model and Bilinear Interpolation to get the final precise rectified images. In this step, a reference image is required and some ground control points (GCPs) are selected. Experiments showed that the final rectified images are satisfactory, and that our two-step rectification algorithm is very effective.
