基于矢量量化的高效BTC图像编码算法

作为一种有损图像编码技术,块截短编码算法（BTC）的计算量较少,速度快,有较好的信道容错力,重建图像质量较高。然而,标准BTC算法的主要缺点是其压缩比特率比其他基于块图像编码的算法（如变换编码和矢量量化）高。为了降低比特率,提出了几种有效的BTC算法,还提出了一种简单的查表算法对每块的BTC量化数据编码,另外还引入了矢量量化技术以减少对位平面编码的比特数。为了减少由改进算法引入的额外失真,在每种提出的算法中,采用最优阈值而不用平均值作为量化阈值。     

Lossless data hiding for color images based on block truncation coding

In this paper, we present a novel, reversible steganographic method, which can reconstruct an original image effectively after extracting the embedded secret data. The proposed reversible hiding method aims at BTC (block truncation coding)-compressed color images. Conventionally, each block of a color image compressed by BTC requires three bitmaps and three pairs of quantization levels for reconstruction. In order to improve the compression rate, a genetic algorithm (GA) is applied to find an approximate optimal common bitmap to replace the original three. The secret data then are embedded in the common bitmap and the quantization levels of each block use the properties of side matching and the order of these quantization levels to achieve reversibility. The experimental results demonstrate that the proposed method is practical for BTC-compressed color images and can embed more than three bits in each BTC-encoded block on average.     

一种非线性图象矢量量化的重建方法——梯度中值算子法

针对图象矢量量化编码复杂度高，并存在方块效应等缺点，本文基于亚－Ｎｙｑｕｉｓｔ采样方式构成训练码矢，并提出了一种新的重建图象方法——梯度中值算子法，它不仅降低了运算的复杂度，而且在插值的同时改善了图象的质量。实验结果表明，在压缩比为０．３１２５ｂｉｔ／ｐｉｘｅｌ下可取得较高的信噪比（ＲＳＮ）和较好的视觉效果     

面向高效率视频编码标准的快速码率控制算法

在详细分析高效率视频编码标准HEVC现有码率控制算法JCTVC-H0213的基础上,提出了一种快速有效的改进算法。该算法根据前一帧的目标比特和实际比特的比值来确定当前帧量化参数。在恒定比特率情况下将该算法与JCTVC-H0213进行比较,实验结果表明该算法对于大部分内容特性序列峰值信噪比变化值提高2.4dB以上,达到了很好的码率控制效果;并且该算法非常简单,适用于实时的码率控制应用。     

基于非局部低秩和自适应量化约束先验的HEVC后处理算法

经高效视频编解码标准HEVC压缩后的视频在高压缩比、低码率的情况下存在明显的压缩效应。针对该问题,提出了一种基于非局部低秩(Non-local Low-rank,NLLR)和自适应量化约束(Adaptive Quantization Constraint,AQC)先验的HEVC后处理算法。该算法首先构造在最大后验概率框架下的优化问题,然后利用解码后的压缩视频和量化参数QP获取非局部低秩和自适应量化约束先验信息,最后利用split-Bregman迭代算法来解决所提的优化问题,从而有效去除压缩效应,提升重建视频质量。其中,非局部低秩先验通过构建基于相似块聚类的非局部低秩模型来获得;自适应量化约束先验通过联合不同量化参数QP下的约束特性与视频的DCT域块活动性来获得。实验结果表明,在同等码率的情况下,与HEVC标准相比,所提算法在帧内编码模式下可以达到平均0.2597 dB的PSNR提升,在帧间编码模式下可以达到平均0.2828 dB的PSNR提升。     

一种基于重构性深度网络的MELP语音编码改进算法

为了提高深度模型的编码重构性能,本文为传统对比散度(Contrastive divergence,CD)添加了基于交叉熵的重构误差约束。利用改进后的算法训练了重构性深度自编码机(Reconstructive deep auto encoder,RDAE),并用RDAE替换混合激励线性预测编码(Mixed excitation linear prediction, MELP)语音编码器中 LSF参数的矢量量化方法。测试结果表明,改进后的算法在损失一定模型似然度的条件下获得了重构性能的提升,当RDAE隐藏层结点设为19　bit时,本文方法所测得的加权LSF距离、重构语音质量、谱失真指标在训练集和测试集上均优于25　bit矢量量化方法,即利用本文方法改进的MELP编码器,在不降低语音质量的条件下,可将MELP编码速率从2.4　kb/s降低至2.1　kb/s,编码速率降低了12.5%。     

Adaptive and dynamic multi-grouping scheme for absolute moment block truncation coding

Image compression technique is widely used in multimedia signal processing. As a conventional lossy compression technique, block truncation coding (BTC) deserves further improvements to enhance its performance of compression. The improvements of BTC mainly focus on: 1) enhancing the quality of reconstructed image and 2) decreasing the bit rate. In this paper, an adaptive and dynamic multi-grouping scheme is proposed for the absolute moment block truncation coding (AMBTC), which is mainly based on an optimized grouping mechanism with the adaptive threshold setting according to the complexity of image blocks. Besides, the values of the reconstruction levels are replaced by their compressed difference values in order to decrease the bit rate. Experimental results demonstrate that the proposed scheme can enhance the compression performance of AMBTC effectively.

     

An oblivious fragile watermarking scheme for images utilizing edge transitions in BTC bitmaps

As one of the famous block-based image coding schemes,block truncation coding(BTC) has been also applied in digital watermarking.Previous BTC-based watermarking or hiding schemes usually embed secret data by modifying the BTC encoding stage or BTC-compressed data,obtaining the watermarked image with poorer quality than the BTC-compressed version.This paper presents a new oblivious image watermarking scheme by exploiting BTC bitmaps.Unlike the traditional schemes,our approach does not really perform the BTC compression on images during the embedding process but utilizes the parity of the number of horizontal edge transitions in each BTC bitmap to guide the watermark embedding and extraction processes.The embedding process starts by partitioning the original cover image into non-overlapping 4×4 blocks and performing BTC on each block to obtain its BTC bitmap.One watermark bit is embedded in each block by modifying at most three pixel values in the block to make sure that the parity of the number of horizontal edge transitions in the bitmap of the modified block is equal to the embedded watermark bit.In the extraction stage,the suspicious image is first partitioned into non-overlapping 4×4 blocks and BTC is performed on each block to obtain its bitmap.Then,by checking the parity of the number of horizontal edge transitions in the bitmap,we can extract one watermark bit in each block.Experimental results demonstrate that the proposed watermarking scheme is fragile to various image processing operations while keeping the transparency very well.     

基于纹理复杂度和运动方向的3D视频水印算法

3D视频的编码方法在HEVC（high efficiency video coding）的基础上有了较大的改动.针对这些改变,提出了一种基于纹理复杂度和运动方向的3D视频水印算法.首先,采用灰度共生矩阵和梯度矩阵分别对I帧中CTU（coding tree unit）和8 CU×8 CU（coding unit）纹理复杂度进行统计分析,将分析结果作为阈值选出P帧和B帧中纹理复杂度较高的8 CU×8 CU;其次,选择帧间预测模式为对称分割的块作为嵌入块,结合同位纹理块的运动矢量分量大小来确定嵌入的水印;最后,通过调制当前块运动矢量的搜索范围或修改区域最优运动矢量的大小来嵌入水印.本算法对不同视频序列平均嵌入容量达到613字节/帧,在不同QPs（quantization parameters）（25、30、35、40）值的重编码攻击下的平均误码率为13.31%,PSNR值平均下降仅为0.007 5 dB,对合成视点质量几乎没有影响,同时鲁棒性好.     

Lapped transform‐based codec for frame‐memory reduction in super‐quality LCD overdrive

Abstract— Overdrive is commonly used to reduce the liquid‐crystal response time and motion blur in liquid‐crystal displays (LCDs). However, overdrive requires a large frame memory in order to store the previous frame for reference. In this paper, a lapped transform‐based codec (LTC) is proposed to reduce the frame memory needed for LCD overdrive. In the latest literature, a directional prediction‐based codec (DPC) employs eight directional predictions, which takes up a large percentage of the computational complexity of the codec and does not consider the de‐correlation of the inter‐blocks. Therefore, the LTC first uses the lapped transform to decompose the correlation of the inter‐blocks in the YUV color space. A hadamard transform is then used for energy compaction. The reordered coefficients are pre‐quantized and encoded using the proposed adaptive bit‐plane coding (ABPC) method for a simple hardware implementation. The simulation results show that the proposed LTC significantly improves the DPC in both subjective and objective performance and outperforms the block truncation coding (BTC) and adaptive multi‐level BTC (AM‐BTC), which have been described in the literature.     

Tamper detection and image recovery for BTC-compressed images

To ensure the integrity of images compressed using block truncation coding (BTC), a tamper detection and image recovery scheme is proposed in this paper. In this scheme, the size of the authentication data can be adaptively selected according to the user’s requirement. The authentication data is embedded in the value differences of the quantization levels in each BTC-compressed image block. Multiple copies of the recovery data are embedded into the bit maps of the smooth blocks. Experimental results show that the proposed scheme performs well in terms of detection precision and the embedded image quality. Meanwhile, the tampered areas can be roughly recovered by using the proposed scheme.     

哈夫曼编码乘积量化的图像哈希检索方法

目的 基于哈希编码的检索方法是图像检索领域中的经典方法。其原理是将原始空间中相似的图片经哈希函数投影、量化后,在汉明空间中得到相近的哈希码。此类方法一般包括两个过程：投影和量化。投影过程大多采用主成分分析法对原始数据进行降维,但不同方法的量化过程差异较大。对于信息量不均衡的数据,传统的图像哈希检索方法采用等长固定编码位数量化的方式,导致出现低编码效率和低量化精度等问题。为此,本文提出基于哈夫曼编码的乘积量化方法。方法 首先,利用乘积量化法对降维后的数据进行量化,以便较好地保持数据在原始空间中的分布情况。然后,采用子空间方差作为衡量信息量的标准,并以此作为编码位数分配的依据。最后,借助于哈夫曼树,给方差大的子空间分配更多的编码位数。结果 在常用公开数据集MNIST、NUS-WIDE和22K LabelMe上进行实验验证,与原始的乘积量化方法相比,所提出方法能平均降低49%的量化误差,并提高19%的平均准确率。在数据集MNIST上,与同类方法的变换编码方法（TC）进行对比,比较了从32 bit到256 bit编码时的训练时间,本文方法的训练时间能够平均缩短22.5 s。结论 本文提出了一种基于多位编码乘积量化的哈希方法,该方法提高了哈希编码的效率和量化精度,在平均准确率、召回率等性能上优于其他同类算法,可以有效地应用到图像检索相关领域。     

一种用于海量图像多分辨率局部重建的压缩算法

论文通过对Shaprio嵌入式小波系数零树编码方法的误差进行分析,针对海量图像数据量庞大的特点,提出了一种便于海量图像局部重建的量化算法。在该算法的基础上,结合位平面编码和游程编码,得到了比较高的压缩比。实验证明这种方法简单有效,易于编程实现,在对相关项目所涉及的海量图像进行处理过程中取得了很好的效果,是一种行之有效的压缩方法。     

5G抽头延迟模型中极化译码后处理方案

考虑5G协议中的抽头延迟信道模型,针对分段循环冗余比特自适应（SCAD-SCL）极化译码算法未考虑信噪比对接收序列对数似然比影响的问题,提出SCAD-SCL的后处理（PSCAD-SCL）译码算法。通过仿真确定用于提高可靠性设定的常数和不可靠对数似然比的数目,翻转并适当放大不可靠的对数似然比值,从而提高误比特率性能。仿真结果表明,PSCAD-SCL的误比特率在信噪比1~3 dB优于原有SCAD-SCL译码算法。在误比特率为10-4时PSCAD-SCL较SCAD-SCL算法性能提升了约0.3 dB,提出算法能够提升5G传输性能。     

一种低码率彩色人脸图像编码方法

提出了一种适合低码率信道传输的嵌入式彩色人脸图像编码方法,不仅把握了人脸图像的特点,而且充分利用了彩色图像小波变换后的多种相关性。该方法将联合色彩分量矢量量化和零块、零树编码方法巧妙结合,把彩色图像的三个色彩分量单独扫描,单独编码的过程简化为联合扫描和联合编码的过程。实验结果表明该算法在低码率下,相比JPEG,获得了在视觉效果和峰值信噪比方面更好的恢复图像质量。     

A 6:1 image‐compression method using directional prediction for LCD overdrive

Abstract— An overdrive technology was developed and is widely used to diminish motion blur in LCDs. To store a previous frame in the overdrive operation in a limited‐sized memory, simple image‐compression techniques are required. By considering the strong correlation of nearby pixels in natural images, a new 6:1 color‐image‐compression method based on directional prediction is proposed. Different from the directional prediction of intra‐coding in H.264/AVC, the predictable direction is determined beforehand to minimize the computation complexity. A simple content‐adaptive quantization and bit‐streaming method, which preserves image details and is free from blocking artifacts, is also proposed. The experimental results show that our proposed method outperforms the vector quantization block truncation coding method with an average 3‐dB peak signal‐to‐noise ratio (PSNR) as well as the subjective quality in terms of blocking artifacts.     

对流层散射信道中5G-NR LDPC码性能

由于目前衰落信道,特别是时变信道非线性环境下的研究极少,因此研究对流层散射信道应用场景下的5G-NR LP-DC编译码性能.5G-NR LDPC码是一种准循环LDPC码,其通过基矩阵构造,可以支持多种码率和码长,便于实现速率自适应.本文提出一种改进的分层归一化最小和译码算法,利用对归一化因子的修正,使该算法更适合应用于...     

VQ indexes compression and information hiding using hybrid lossless index coding

For the compression of memoryless vector quantization (VQ), most of the lossless index coding algorithms are not suitable for various test images. As a result, we present a hybrid dynamic tree-coding scheme (DTCS) and modified search order coding scheme (MSOC) to re-encode the output index map efficiently without causing any extra coding distortion. The main idea behind this scheme is that the adjacent left and upper around the current processed block usually provide more useful information than its adjacent left-upper and right-upper block, thus we employ two different coding methods according to their corresponding left or upper spatial relations. In addition, we applied the HLIC method to the information hiding. The proposed method does not modify the contents of the secret data and the compressed image. Experimental results show that the newly proposed algorithm achieves significant reduction of bit rate compared to the other lossless index coding schemes for various test images and different codebook sizes. The proposed information hiding scheme can hide a huge amount of information in the index map of an image and allows complete reconstruction of the indexes of the image.     

A sequential initialization technique for vector quantizer design

Vector quantization has been used in compressing both speech and image data. In theory, better performance can always be achieved by coding vectors instead of scalars. However, actual results depend upon the proper design of the quantizer. Vector quantizer design typically employs an algorithm such as the K-means algorithm or the Linde Buzo Gray algorithm in which the initialization affects the design cost (convergence rate) and the achievable performance (quantization error). After reviewing several current initialization techniques, a sequential initialization method called Error Function Initialization is presented. In this method, the seeds are chosen one at a time by attempting to maximize the step-wise reduction in the quantization error. Experimental results show that this technique yields faster convergence and smaller quantization errors. For real time applications, the technique could be used to design sub-optimal vector quantizers.     

Modified block truncation coding for image compression

This paper presents a modified block truncation coding scheme for the compression of images. We first design a set of binary edge patterns, which are visually significant, to approximate the bit plane of an image block. An interblock coding scheme, utilizing the spatial correlation between neighboring blocks, is then developed for coding of the sample mean and standard deviation of a block. Simulation results indicate that the bit rate is significantly reduced without introducing noticeable degradation in the reconstructed images.