L/M-fold image resizing in block-DCT domain using symmetric convolution

Image resizing is to change an image size by upsampling or downsampling of a digital image. Most still images and video frames on digital media are given in a compressed domain. Image resizing of a compressed image can be performed in the spatial domain via decompression and recompression. In general, resizing of a compressed image in a compressed domain is much faster than that in the spatial domain. We propose a novel approach to resize images with L/M resizing ratio in the discrete cosine transform (DCT) domain, which exploits the multiplication-convolution property of DCT (multiplication in the spatial domain corresponds to symmetric convolution in the DCT domain). When an image is given in terms of its 8/spl times/8 block-DCT coefficients, its resized image is also obtained in 8/spl times/8 block-DCT coefficients. The proposed approach is computationally fast and produces visually fine images with high PSNR.     

Arbitrary resizing of images in DCT space

Using the spatial relationship of the block discrete cosine transform (DCT) coefficients and subband approximations, algorithms for image halving and doubling operations are presented. The computational steps identified in the process provide a general framework for image resizing operations. Some of the previously reported image halving and doubling algorithms are shown to be special cases. The proposed approach is general enough to accommodate resizing operations with arbitrary factors, namely with integral and rational factors. The application of these methods to the conversion in the compressed domain of images (video frames) from one format to another is demonstrated.     

Arbitrary resizing of images in DCT domain using lapped transforms

An arbitrary L/M-fold image resizing method using lapped transforms is presented. The resizing operation is carried out in the lapped transform domain, by converting the images in the discrete cosine transform (DCT) domain into those in the lapped transform domain and vice versa. The proposed method provides visually fine images, while reducing the blocking effect to a very low level for images compressed at low bit rates.     

基于重组DCT系数子带能量直方图的图像检索

现在许多图像采用JPEG格式存储,检索这些图像通常要先解压缩,然后提取基于像素域的特征矢量进行图像检索。己有文献提出直接在DCT域进行图像检索的方法,这样可以降低检索的时间复杂度。本文提出对JPEG图像的DCT系数利用多分辨率小波变换的形式进行重组,对整个数据库中所有图像的DCT系数重组得到的若干子带,分别建立子带能量直方图,而后采用Morton顺序建立每幅图像的索引,并采用变形B树结构组织图像数据库用于图像检索。     

Fast Inverse Motion Compensation Algorithms for MPEG and for Partial DCT Information

In prior work, we developed a fast inverse motion compensation method that can be implemented directly on the DCT domain representation derived from the compressed bitstreams conforming to MPEG, H.261, and H.263 standards. That work was restricted to compressed-domain representations wherein the motion-vectors have integer pel accuracy. Here, we extend this work to fractional-pel accurate motion-vectors. We also extend the prior work to speed up the inverse motion compensation process in the DCT domain by explicitly exploiting the sparseness of the DCT domain representation. Using partial DCT information, we show that the DCT domain method has substantially lower operation count than the conventional spatial domain approach which requires decompression followed by inverse motion-compensation.     

A new texture generation method based on pseudo-DCT coefficients.

In this paper, a new method for generating different texture images is presented. This method involves a simple transform from a certain one-dimensional (1-D) signal to an expected two-dimensional (2-D) image. Unlike traditional methods, the input signal is generated by a simple 1-D function in our work instead of a sample texture. We first transform the 1-D input signal into frequency domain using fast Fourier transform. Based on the sufficient analysis in 2-D discrete cosine transform (DCT) domain, where each of the coefficients expresses a texture feature in a certain direction, the 2-D pseudo-DCT coefficients are then constructed by appropriately rearranging the Fourier coefficients in terms of their frequency components. Finally, the corresponding texture image can be produced by 2-D inverse DCT algorithm. We applied the proposed method to generate several stochastic textures (i.e., cloud, illumination, and sand), and several structural texture images. Experimental results indicate the good performance of the proposed method.     

Wavelet to DCT transcoding in transform domain

Wavelet to DCT transcoding provides inter-operability between standards using the two transforms for encoding. Transcoding in transform domain avoids inverse transform and re-transform operations and saves computation. In this paper, we propose new algorithms for transcoding wavelet coefficients to block DCT coefficients. In the first step, the wavelet coefficients are transformed into upsampled DCT coefficients. Subsequently, these trans-formed coefficients are synthesized in the block DCT space for transcoding. The proposed approach restricts all operations in the DCT domain that makes filtering involved in the synthesis process computationally efficient. The proposed technique could be used by the block DCT based services when the input is available as wavelet coefficients.     

Fast DCT domain filtering using the DCT and the DST

A method for efficient spatial domain filtering, directly in the discrete cosine transform (DCT) domain, is developed and proposed. It consists of using the discrete sine transform (DST) and the DCT for transform-domain processing on the in JPEG basis of the previously derived convolution-multiplication properties of discrete trigonometric transforms. The proposed scheme requires neither zero padding of the input data nor kernel symmetry. It is demonstrated that, in typical applications, the proposed algorithm is significantly more efficient than the conventional filtered spatial domain and earlier proposed DCT domain methods. The proposed method is applicable to any DCT-based image compression standard, such as JPEG, MPEG, and H.261.     

一种快速的基于稀疏表示和非下采样轮廓波变换的图像融合算法

为了提高图像融合的效率和质量,该文提出一种基于快速非下采样轮廓波变换(NSCT)和4方向稀疏表示的图像融合算法。该方法首先对源图像进行快速NSCT分解,生成一系列低通和高通子带。对于低频子带,利用自适应生成的DCT过完备字典进行快速的4方向稀疏表示和系数融合;对于高频子带,则利用高斯加权区域能量最大的融合规则进行系数融合。快速NSCT将传统NSCT的树形滤波结构转变为多通道滤波结构,能成倍提高分解效率;快速的稀疏融合则抛弃了传统的滑动窗口方法,以水平、垂直、对角线4个方向进行稀疏表示和稀疏融合,进一步提高算法效率。实验结果表明,提出的快速算法能在不影响融合质量的条件下将算法效率提高近20倍。     

Saliency detection in the compressed domain for adaptive image retargeting

Saliency detection plays important roles in many image processing applications, such as regions of interest extraction and image resizing. Existing saliency detection models are built in the uncompressed domain. Since most images over Internet are typically stored in the compressed domain such as joint photographic experts group (JPEG), we propose a novel saliency detection model in the compressed domain in this paper. The intensity, color, and texture features of the image are extracted from discrete cosine transform (DCT) coefficients in the JPEG bit-stream. Saliency value of each DCT block is obtained based on the Hausdorff distance calculation and feature map fusion. Based on the proposed saliency detection model, we further design an adaptive image retargeting algorithm in the compressed domain. The proposed image retargeting algorithm utilizes multioperator operation comprised of the block-based seam carving and the image scaling to resize images. A new definition of texture homogeneity is given to determine the amount of removal block-based seams. Thanks to the directly derived accurate saliency information from the compressed domain, the proposed image retargeting algorithm effectively preserves the visually important regions for images, efficiently removes the less crucial regions, and therefore significantly outperforms the relevant state-of-the-art algorithms, as demonstrated with the in-depth analysis in the extensive experiments.     

DCT域中MPEG7主色描述符的提取

该文在MPEG7的基础上提出了DCT域内直接提取主色描述符的新方法。这种方法节省了对图像的解压缩的过程,因而大大的提高了对于压缩图像进行特征提取的速度和效果。作为整个箅法的一部分,一种自动阈值提取的算法也在该文中给予了描述。这种方法可以减少因人为设定经验阈值而带来的不确定性,使算法更具鲁棒性。对比检索试验结果也说明本算法是一个高速有效的算法。新算法主要用于压缩图像库或互联网上的相似检索。     

一种基于Wyner-Ziv结构的贝尔模板图像编码方法

 该文提出一种基于Wyner-Ziv(WZ)结构的四通道贝尔模板图像分布式编解码方法。在编码端对贝尔模板图像进行结构分离转换,形成4个分量图像并分别执行离散余弦变换,依据拉格朗日代价函数的收敛性,利用Lloyd迭代算法设计了全局优化的量化器,采用Slepian-Wolf(SW)信道编码方法对各分量变换系数的量化输出进行独立编码,在解码端利用亮度分量作为边信息,联合解码重构贝尔模板图像。实验表明,在高速率情况下,其率失真性能得到较好的改善。     

A direct computation of DCT coefficients for a signal block takenfrom two adjacent blocks

A direct computational algorithm for obtaining the DCT (discrete cosine transform) coefficients of a signal block taken from two adjacent blocks is proposed. This algorithm reduces the number of both multiplications and additions/subtractions compared to the traditional method, which requires inverse transforms of two received coefficient blocks followed by a forward transform. The proposed algorithm is efficient and useful for speech coding and image processing, especially for real-time applications     

Block DCT to wavelet transcoding in transform domain

The Discrete Cosine Transform (DCT) to wavelet transcoding provides input for several wavelet-based post-processing techniques of the DCT-coded image/video signals. Transcoding in domain transform avoids inverse transform and retransform operations and saves computation. In this paper, we propose a new technique for transcoding the DCT blocks to wavelet coefficients directly in the transform domain. We perform filtering, IDCT and downsampling operations in a single combined step. The proposed technique achieves the same computational result as that of a spatial domain technique. The transcoding matrices used in the proposed technique are found to satisfy certain symmetric and sparse properties, which are exploited to reduce the computational cost. As the number of zeros in the DCT coefficients is significantly higher compared to the spatial domain, computational cost reduces significantly. Also, with the proposed technique, it is possible to speedup the operation by ignoring some elements in the filtering matrices whose magnitudes are smaller than a threshold value. We demonstrate the application of the proposed transcoding for deblocking of the DCT-coded images in wavelet domain.     

基于独立分量分析的压缩域彩色图像融合

提出了一种基于独立分量分析的压缩域彩色图像融合方法,其基本思想是将源彩色图像进行IHS变换,将两幅图像的亮度分量/进行小波分解,对分解后的高频系数采用由独立分量分析得出的变换矩阵进行变换,然后对变换系数采用局部方差法进行融合,而对小波分解后的低频子图像分成若干N×N的图像块,分别对每个图像块做二维DCT变换,最后对融合的系数进行反变换得到融合结果,并用客观评价标准对结果进行了定量的分析.实验结果表明:该方法在提高空间信息的基础上,较少地降低了颜色失真.     

遥感图像自适应分层量化的快速DCT压缩法

依据遥感图像的频谱特性,提出一种自适应分层量化的快速DCT图像压缩算法,在对原始图像快速DCT之后,根据图像频谱特性自适应修正JPEG量化表,再用新量化表分层量化DCT系数。真实遥感图像压缩实验表明,在同等压缩比下,提出的方法比标准JPEG方法速度快,且峰值信噪比增加1～2dB,并能实现嵌入式码流图像压缩。     

Image extraction in DCT domain

More and more digital images are being stored in compressed formats, among which the format using discrete cosine transform (DCT) coefficients is widely adopted (JPEG, MPEG, H.263 etc). To exploit those successful image processing techniques developed in the pixel domain, the authors propose a fast image extraction algorithm to allow images to be extracted directly from compressed DCT coefficients without full decompression. In the proposed technique the extracted images retain quality comparable with that of fully decoded images. However, the computing cost and the storage expense incurred by the proposed algorithm are very much lower than the costs of full decompression. The experiments also demonstrate that the proposed algorithm has tremendous potential in that all existing image processing techniques developed in the pixel domain can be directly applied to compressed images, without involving full decompression.     

Content Authentication Based on JPEG-to-JPEG Watermarking

A content authentication technique based on JPEG-to-JPEG watermarking is proposed in this paper. In this technique, each 88 block in a JPEG compressed image is first processed by entropy decoding, and then the quantized discrete cosine transform (DCT) is applied to generate DCT coefficients: one DC coefficient and 63 AC coefficients in frequency coefficients. The DCT AC coefficients are used to form zero planes in which the watermark is embedded by a chaotic map. In this way, the watermark information is embedded into JPEG compressed domain, and the output watermarked image is still a JPEG format. The proposed method is especially applicable to content authentication of JPEG image since the quantized coefficients are modified for embedding the watermark and the chaotic system possesses an important property with the high sensitivity on initial values. Experimental results show that the tamper regions are localized accurately when the watermarked JPEG image is maliciously tampered.     

基于二维最大边缘准则和DCT的人脸识别方法

提出了一种基于离散余弦变换(DCT)和二维最大边缘准则(2DMMC)的2DDM特征提取算法,证明了2DMMC可以直接应用于DCT域,利用欧氏距离测度进行分类的结果与在空域中进行得到的结果完全相同。2DMMC方法可直接应用于基于DCT压缩的JPEG格式的图像。在ORL和Yale人脸数据库上的实验结果表明,在空域2DMMC的识别率高于2DPCA和2DLDA,2DDM的识别率又高于2DMMC,而且2DDM的耗时要低于2DMMC。     

基于FPGA的高分辨率图像DCT域增强

为了提高高分辨率图像的质量,实现快速的图像增强算法,提出在离散余弦变换（DCT）的对比度测度下,通过DCT矩阵中不同频率的系数关系对DCT系数块进行分类,对不同类型的系数块做不同强度的自适应增强算法,并在FPGA上得到实现。提出的方法在不影响原始图像压缩性能的情况下有效地增强了图像明亮或黑暗区域的细节,同时减少了因图像增强而带来的压缩图像块效应。给出算法原理及在FPGA上的具体实现方法,并给出了实验结果。结果表明,该算法在改善图像主、客观质量方面和运算效率上都能够达到较好的效果。