基于小波变换与稀疏傅里叶变换相结合的光场重构方法

发布时间：2018-07-17 03:43

【摘要】：随着计算机图形学和计算机视觉技术的发展,光场开始进入人们的视线并被迅速应用于各个领域.然而光场的获取需要大量的图像,具有数据量大,获取成本高等特点,因此学者们越来越关注如何利用少量的光场数据获取整个光场这一问题,并且做出了大量的工作.针对上述问题,本文将小波变换与稀疏傅里叶变换相结合,利用光场在角度域的稀疏性提出一种新的光场重构方法.首先,利用小波变换多分辨率分析的特点,通过小波变换将原始图像分解为多个不同频率的子图像;然后分别对每个子图像通过傅里叶切片定理恢复其频率位置,从而可以分别得到它们的二维角度谱;最后将每个子图像的二维角度谱合并,进行小波逆变换获得整个光场.本文方法利用小波变换将原图像分解为多个不同频率的子图像分别同时处理,不仅降低了算法的复杂度,大大减少了算法的运行时间,为光场的广泛应用提供了条件,而且相比于单独运用稀疏傅里叶算法重构,本方法有效地抑制了窗口效应,使重构结果更加准确.此外,本文方法将高频信息和低频信息分开重构,可以有效地改善并网恢复中小频率丢失的问题,进一步改进重构结果.最后通过仿真验证了算法的有效性.
[Abstract]:With the development of computer graphics and computer vision technology, light field has been applied to many fields. However, the acquisition of light field requires a large number of images, which has the characteristics of large amount of data and high cost. Therefore, scholars pay more and more attention to the problem of how to obtain the whole light field using a small amount of light field data, and have done a lot of work. Aiming at the above problems, a new method of light field reconstruction is proposed by combining wavelet transform with sparse Fourier transform and utilizing the sparsity of light field in angle domain. Firstly, the original image is decomposed into several sub-images with different frequencies by wavelet transform, and then the frequency position of each sub-image is restored by Fourier slicing theorem. Finally, the two-dimensional angle spectrum of each sub-image is combined and the whole light field is obtained by inverse wavelet transform. In this paper, wavelet transform is used to decompose the original image into several sub-images with different frequencies, which not only reduces the complexity of the algorithm, but also greatly reduces the running time of the algorithm, which provides the conditions for the wide application of the light field. Compared with the sparse Fourier algorithm alone, this method can effectively suppress the window effect and make the reconstruction results more accurate. In addition, the reconstruction of high frequency information and low frequency information by this method can effectively improve the problem of recovering small and medium frequency loss in grid-connected restoration, and further improve the reconstruction results. Finally, the effectiveness of the algorithm is verified by simulation.
【作者单位】：杭州电子科技大学自动化学院;浙江大学电气工程学院;
【基金】：国家自然科学基金重点项目(No.61333011);国家自然科学基金(No.61371064,No.61273075,No.61503206)
【分类号】：TP391.41

【相似文献】