Hou Biao

Segmentation and recognition of bridges in high resolution SAR images

Based on analyzing the features of SAR images, this paper presents a novel algorithm for automatic segmentation and recognition of bridges in high-resolution SAR images according to the idea of detecting the targets along the edges of the image. At the same time, wavelet analysis is successfully used in prefiltering and denoising; good recognition results are thus obtained.

Image denoising based on ridgelet

Image denoising is an important step in the pre-processing of images; the noisy images bears different characteristics. For an anisotropic image, wavelets lose their effects on singularity detection because discontinuities across edges are spatially distributed. Based on the idea of ridgelet, this paper proposes a new denoising method of anisotropic noisy images combining the dyadic wavelet; good results are thus obtained.

An Algorithm for SAR Image Embedded Compression Based on Wavelet Transform

A novel algorithm for voice conversion is proposed in this paper. The mapping function of spectral vectors of the source and target speakers is calculated by the canonical correlation analysis (CCA) estimation based on Gaussian mixture models. Since the spectral envelope feature remains a majority of second order statistical information contained in speech after linear prediction (LPC) analysis, the CCA method is more suitable for spectral conversion than MMSE because CCA explicitly considers the variance of each component of the spectral vectors during conversion procedure. Both subjective and objective evaluations are conducted. The experimental results demonstrate that the proposed scheme can achieve better performance than the previous method which uses MMSE estimation criterion.Synthetic aperture radar (SAR) image compression is important in image transmission and archiving. In this paper, SAR image compression using embedded zerotree wavelets algorithm, based on discrete wavelet transform (DWT), is researched. Aiming at special characteristics of SAR image such as speckle noise, the author added a denoise step in the flow of compression, trying to do some meaningful attempt in SAR image compression. Experiments carried out show that the improvement can reduce the speckle noise and improve the computation precision and time.

Unsupervised image segmentation based on the anisotropic texture information

Based on the anisotropic characteristic of brushlet, a new feature named directional texture histogram in brushlet domain is presented, which represents the local anisotropic information in image. A novel unsupervised image segmentation method via local directional texture histogram in the brushlet domain is developed. The segmentation results of synthetic mosaics, aerial photo and synthetic aperture radar (SAR) image show that our method represents better performance and its error probability of the synthetic mosaics is lower than the wavelet-based method.

Subband block coding of images through M-band wavelet transform

A subband block coding method is proposed, by exploiting the frequency characteristic of the M-band wavelet transform of images. Two consequent problems are efficiently solved, i.e. (a) how to select proper subbands for reconstruction, and (b) how to determine the size of the block of selected subbands. The bound effect is efficiently removed. Subband block coded images are presented along with their PSNR and compression ratios.