Jiulong Zhang

Face Recognition Based on Curvefaces

A new method called curvefaces was firstly presented for face recognition, which is based on curvelet transform. Curvelet is the latest multiscale geometric analysis tool. Contrast to wavelet transform, curvelet transform directly takes edges as the basic representation elements and is anisotropic with strong direction. It is a multiresolution, band pass and directional function analysis method which is useful to represent the image edges and the curved singularities in images more efficiently. It yields a more sparse representation of the image than wavelet and ridgelet transform. In face recognition, the curvelet coefficients can better represent the main features of the faces. The support vector machine (SVM) can then be used to classify the images. SVM is based on the statistical learning theory and is especially valid for small sample set and can get high recognition rate. Multi-class SVM is employed in this paper. The simulation shows that the proposed method is better than wavelet based method.

Digital Image Watermark Algorithm in the Curvelet Domain

Watermark is a new effective digital copyright protection method and data security technology. In this paper, a digital image watermark algorithm operating in the curvelet domain is proposed. Unequally-Spaced Fast Fourier Transforms (USFFT) is employed to implement Fast Discrete Curvelet Transforms (FDCT ) to transform the digital image to the curvelet domain. The following three factors are considered when embedding the watermark in the curvelet domain: (1) the scale distribution of the curvelet coefficients; (2) the variation of high frequency components and low frequency components have different effects on HVS; (3) the impact of the variation of the curvelet coefficients on the robustness of the watermark. Combining these factors with the basic watermark algorithm, we choose the proper position to embed the watermark. The experimental results demonstrate a fine unification of robustness and invisibility and shows advantages over watermark algorithm in the ridgelet domain concerning the two aspects.

A comparative study of wavelet and Curvelet transform for face recognition

Wavelet transform had been a popular feature extraction tool for face recognition, and much work has been done regarding the choice of wavelet subbands in presence of variation in expression, illumination and pose. However the using of curvelet transform, which is effective in representing curves, and the performance analysis have not been studied. We propose to employ curvelet for facial feature extraction, and perform a thorough comparison against wavelet transform; especially, the orientation of curvelet is analyzed. Experiments show that for expression changes, the small scale coefficients of curvelet outperform the high frequency coefficients of wavelet to a large degree. Whilst for illumination changes, the small scale coefficients of curvelet transform are robust, though the large scale coefficients of both transform are likely influenced. The reason behind the advantages of curvelet lies in its abilities of spatial localization and orientation sensitivity, thus the experiments and theoretical analysis coincide.

Performance Enhancement of DS/SS System Using Hybrid Chaotic Spreading Sequence

In this paper, a chaotic sequence generator based on logistic map and an interleaver is proposed and used to generate spreading sequence. The generated chaotic sequence is balanced itself and has excellent correlation property and sidelobe property. When the sequence is applied to direct sequence/spread spectrum (DS/SS) system, multiple access interference (MAI) can be mitigated by using the spreading sequences. The system capability, performance can be enhanced, and the information can be effectively encrypted. The bit-error rate (BER) performance of spreading sequence from the proposed chaotic sequence generator is compared with that of the conventional m-sequence and Gold sequence. Results show that the new sequence has a noticeable improvement in BER performance in low signal-to-noise ratio (SNR). Although the chaotic sequence has more excellent correlation property and sidelobe property, it is easy to be implemented.

Automatic Building Extraction from Terrestrial Laser Scanning Data

of point clouds with different local densities, especially in the presence of random noisy points, is still a formidable challenge. In this paper, we present a complete strategy for building extraction from terrestrial laser scanning data. First, a novel segmentation method is proposed to facilitate the task of building extraction. The points are grouped based on the normals and the adjacency relationships. Second, the planar surfaces are recognized from the segmentation results based on the properties of the Gaussian image. Finally, the buildings are extracted from the urban point clouds based on a collection of characteristics of point cloud segments like shape, normal direction and topological relationship. Experimental results demonstrate that the proposed method can be used as a robust way to extract buildings from terrestrial laser scanning data. At the same time, the buildings are decomposed into several patches which lay a good foundation for building reconstruction.

Tree branching reconstruction from unilateral point clouds

Trees are ubiquitous in natural environment and realistic models of tree are also indispensable in computer graphics and virtual reality domains. However, their complexity in geometry and topology make it a great challenge for photo-realistic tree reconstruction. Since tree trunk is the preliminary structure of trees, its modeling is a critical step which plays an important role in tree modeling. Many existing methods focus on the overall resemblance of tree branches but omit the local geometry details. In this paper, we perform unilateral scanning of real-world trees and propose an approach that could reconstruct trees from incomplete point clouds. The core of our method contains four parts: local optimal segmentation of tree branch, skeletal point and lines extraction from unilateral branch, the cross-section construction of tree branch, and final tree branch surface generation. Experimental results demonstrate the effectiveness and robustness of our method which could keep realistic shape of trees.

A Generation Method of Chinese Meticulous Painting Based on Image

Meticulous painting is one of Chinese traditional painting that has more than 2000 years history. We present an automatic technique that generates Chinese meticulous painting from image in this paper. First, we transfer the input image to grayscale, and extract the contours by using the Laplace operator. The contour image is the line-drawing draft of the meticulous painting. Second, we propose a grayscale hierarchical mixed method to produce the first layer ink-dyeing draft. The final ink-dyeing draft is produced by mixing the first layer draft and the second layer, which is created by Gaussian smooth. Then, we transfer the input image to HSV color space, and use the H, S and the grayscale of ink-dyeing draft to get the color draft. Finally, we have the meticulous painting by combining the colorful draft with rice paper texture. Experimental results demonstrate the effectiveness of our method in producing meticulous painting from image.

Ridgelet transform with application in ground penetrating radar processing

Ridgelet is the new development of wavelet dealing with line singularities in signals. In goundpenetrating radar (GPR) signal, the direct wave is a strong noise which has line singularity. In time-space domain, the useful information in GPR signal is badly polluted by strong direct wave so that the underground targets can not be recognized correctly. Ridgelet overcomes the disadvantages of wavelets in dealing with high dimensions that it can effectively represent singularities along lines or hyperplanes. Concerning the characteristics of GPR signals, the ridgelet transform is employed to remove the background noise and separate the direct wave. The simulation results for real GPR signals demonstrate its effectiveness.