摘要
虽然2D Gabor变换在图像处理等很多领域认为是非常有用的时频分析的方法,然而实时应用却因其很高的计算复杂性而受到限制。文中回顾了基于DCT的2D的实值离散Gabor变换,为了有效地和快速地计算实值离散Gabor变换,提出了在临界抽样条件下,二维实值Gabor变换系数求解的块时间递归算法以及由变换系数重建原信号的块时间递归算法,研究了该算法使用并行格型结构的实现方法,并讨论和比较了算法的计算复杂性和优越性,证明了基于DCT的2D实值离散Gabor变换块时间递归算法并行格结构在计算复杂性的高性能。
2D Gabor transforms are considered as an effective time - frequency analysis technique in diverse areas as image processing ; however, its real time applications have been limited due to its high computational complexity. In this paper, firstly the 2D real - valued discrete Gabor transform based on DCT will be reviewed; secondly, block- recursive algorithms for the efficient and fast computation of 2 - D RDGT coefficients and for the fast reconstruction of the original signal from the RDGT coefficients will be developed in critical sampling case ; thirdly, unified parallel lattice structures for the implementation of the algorithms wilt be studied; and finally, the computational complexity and the advantages of the proposed algorithms will be discussed and compared, and also proved its high performance of computation complexity.
出处
《计算机技术与发展》
2009年第2期105-108,共4页
Computer Technology and Development
基金
国家自然科学基金(60572128)
安徽省人才开发基金(20052029)