A Novel Fast and Reduced Redundancy Structure for Multiscale Directional Filter Banks

Abstract

The multiscale directional filter bank (MDFB) improves the radial frequency resolution of the contourlet transform by introducing an additional decomposition in the high-frequency band. The increase in frequency resolution is particularly useful for texture description because of the quasi-periodic property of textures. However, the MDFB needs an extra set of scale and directional decomposition, which is performed on the full image size. The rise in computational complexity is, thus, prominent. In this paper, we develop an efficient implementation framework for the MDFB. In the new framework, directional decomposition on the first two scales is performed prior to the scale decomposition. This allows sharing of directional decomposition among the two scales and, hence, reduces the computational complexity significantly. Based on this framework, two fast implementations of the MDFB are proposed. The first one can maintain the same flexibility in directional selectivity in the first two scales while the other has the same redundancy ratio as the contourlet transform. Experimental results show that the first and the second schemes can reduce the computational time by 33.3%-34.6% and 37.1%-37.5%, respectively, compared to the original MDFB algorithm. Meanwhile, the texture retrieval performance of the proposed algorithms is more or less the same as the original MDFB approach which outperforms the steerable pyramid and the contourlet transform approaches.

Analysis and Modeling of Statistical Properties of FMDFB Subband Coefficients

“…The over completeness and increased frequency resolution of MDFB was found useful in applications like texture characterization and retrieval (Cheng et al, 2007a). Cheng et al (2007b) proposed a fast and reduced redundancy structure for this MDFB (FMDFB). This structure has the same redundancy as PDFB and 33% reduction in computational complexity compared to MDFB.…”

“…The total number of FMDFB directional subband coefficients is the same as the size of the original image because of the critically sampled DFB, and hence no extra computations are introduced by the scale decomposition. The computational complexity of FMDFB (Cheng et al, 2007b) is approximated as…”

“…Fast multiscale directional filter bank is an image representation scheme that is successfully used in feature extraction and image enhancement (Cheng, Law, & Siu, 2007a), ( Cheng, Law, & Siu, 2007b). In our previous work, an image denoising scheme with FMDFB is proposed that works well for additive Gaussian and multiplicative speckle noise removal compared to other conventional multiscale denoising methods such as wavelet 803 and contourlet based denoising (Leavline, Sutha, & Singh, 2014a).…”

Design of FIR Filters for Fast Multiscale Directional Filter Banks

“…Cheng et al proposed a fast and critically sampled structure for MDFB namely as fast multiscale directional filter banks [6] shown in Figure 1 In analysis filter bank structure, low pass filtering and wavelet transform technique are applied for splitting the image into various scales. Use of non aliasing lowpass filters broadens the bandwidth of the finer scale.…”

“…Cheng, et al, [6] proposed a fast structure for this MDFB named as fast multiscale directional filter banks (FMDFB). This filter structure is found suitable for various image processing applications such as image retrieval, feature extraction, and denoising.…”