Search

Abstract: The contourlet transform is an emerging multiscale multidirection image processing technique. It effectively represents smooth curvature details typical of natural images, overcoming a major drawback of the 2-D wavelet transform. Previously, we developed a contourlet image model, that is, the contourlet contextual hidden Markov model (C-CHMM). In this paper, we further develop a multiscale texture segmentation technique based on the C-CHMM. The segmentation method combines a model comparison approach with a multiscale fusion and a neighbor combination process. It also features a neighborhood selection scheme based on smoothed context maps, for both model estimation and neighbor combination. Through a series of segmentation experiments, we examine the effectiveness of the C-CHMM in comparison with closely related models. We also investigate how different context designs affect the segmentation performance. Moreover, we show that the C-CHMM based technique provides improved accuracy in segmenting texture patterns of diversified nature, as compared with popular methods such as the HMTseg and the JMCMS. All these simulation experiments demonstrate the great potential of the C-CHMM for image analysis applications. [Copyright &y& Elsevier]

A novel tunable-quality-factor (tunable-Q) contourlet transform for geometric image representation is proposed. The Laplacian pyramid in original contourlet decomposes a signal into channels that have the same bandwidth on a logarithmic scale, and is not suitable for images with different behavior in frequency domain. We employ a new tunable-Q decomposition defined in the frequency domain by which one can flexibly tune the bandwidth of decomposition channels. With an acceptable redundancy, this tunable-Q contourlet is also anti-aliasing and its basis is sharply localized in the desired area of frequency and spatial domain. Our experiments in nonlinear approximation and denoising show that the contourlet using a better-suitable quality factor can achieve a more promising performance and often outperform wavelets and the previous contourlets both in visual quality and in terms of peak signal-to-noise ratio. [ABSTRACT FROM AUTHOR]