摘要
由于室外场景图像集具有规模巨大、尺度多变等特点,快速精准的多视图立体匹配受计算效率严重制约。为此,提出一种新的快速多视图立体匹配和优化方法。该方法首先基于Plane Sweep框架计算初始匹配代价;然后提出并行Semi-Global算法对匹配代价优化计算深度图;最后使用GPU对图像滤波剔除噪声点。实验结果表明,该方法可高效生成用于三维重建的优质深度图。
The accuracy and effieiency of multi-view stereo matching are seriously hindered by the scale of the image datasets and the different variety of dimension in outdoor scenes. To deal with it, proposes a novel quick multi-view stereo matching method. Firstly, computes raw matching cost by plane sweep framework. Then, proposes a parallel matching cost computing method based on Semi-global algorithm. Finally, uses the fihering to get rid of the image noise with GPU. Experimental results confirm the performance of the proposed method, which can be efficienlly generate high-quality depth map for 3D reconstruction.
出处
《现代计算机》
2018年第1期18-21,共4页
Modern Computer
基金
四川省科技创新苗子工程资助项目(No.2017-003
No.2017-004)
