Parallel optical flow estimation with capturing images in real time
Parallel optical flow estimation with capturing images in real time
复制标题
通过实时捕获图像进行并行光流估计
DOI：
10.1109/iccas.2016.7832308
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
Hiroshi Harada
中科院分区：
文献类型：
--
作者：
Yuta Hamada;Teruo Yamaguchi;Hiroshi Harada
时空分异法具有计算成本低、不需要特征点等优点，是确定运动物体速度最有效的方法之一。该方法有望在实际应用中得到应用。然而，光流的实时估计是困难的，因为它必须等待3帧图像，这是时空分异方法所要求的。实时估计光流的理想程序应该计算每帧运动物体的速度。为了开发这样的程序，我们引入了一种多核编程，可以并行捕获图像并计算运动物体的速度。在测量光流估计的时间时，我们发现特征值的推导时间占了处理时间的大部分。为了减少这一时间，我们改进了计算特征值的方法。此外，为了获得更精确的光流，我们比较了预滤波器对估计光流的影响。
Spatio-temporal differentiation method is one of the most effective methods of determining the velocity of moving objects owing to its low calculation cost, unnecessity of feature points. It is expected that the method can be used in real world. However, it is difficult to estimate optical flow in real time, because it must wait for 3 frames of image that the spatio-temporal differentiation method requires. Ideal program to estimate optical flow in real time should evaluate the velocity of moving objects each frame. In order to develop such program, we introduced a multi-core programming that captures image and evaluates the velocity of moving objects in parallel. In advance, when we measured time to estimate optical flow, it was found that the time for derivation of the eigenvalues accounts for most of the processing time. In order to reduce this time, we improved the method of calculating the eigenvalues. Moreover, to get more accurate optical flow, we compared pre-filters preferable to estimate optical flow.