Location:
Search - quantization
Search list
Description: Compression and decompression VQ=vector quantization for image.
Platform: |
Size: 370688 |
Author: Lusa Elena |
Hits:
Description: 正弦波信号的均匀量化和线性编码
A/D 变换，实数的线性编码
D/A 变换，实数的线形译码-Uniform quantization of sinusoidal signal and linear coding A/D conversion, the linear coding real D/A conversion, real decoding of linear
Platform: |
Size: 1024 |
Author: zhangwei |
Hits:
Description: Gray先生的大作，描述高斯混合矢量量化方法进行说话人识别的方法-Daisaku Mr. Gray describes the Gaussian mixture vector quantization approach to speaker recognition method
Platform: |
Size: 342016 |
Author: 高斯 |
Hits:
Description: 在matlab环境下提取jpeg图像量化表-Extract jpeg image quantization table
Platform: |
Size: 60416 |
Author: leiyu |
Hits:
Description: 均匀PCM量化和u律非均匀量化的对比效果分析实验.-Uniform quantization and u law PCM non-uniform contrast quantitative analysis of experimental results.
Platform: |
Size: 3072 |
Author: tess |
Hits:
Description: H.264图像编码程序对输入像素块X的整数DCT变换和量化-H.264 encoding process the input image pixel block X integer DCT transform and quantization
Platform: |
Size: 2048 |
Author: ydl |
Hits:
Description: 运动估计是视频编码的关键技术，其最基本的原理是利用相邻帧间的时间
相关性，通过预测来减少时间冗余度。在实际编码中，为了节省码率，并不传
输每一帧的全部数据，而是利用运动估计求出每一帧与其预测参考帧之间的差
值。运动估计越准确，差值的分布越趋近与零，差值块的能量越小，经过变换、
量化和熵编码后所产生的码流的比特位率也越少。因此，运动估计搜索的准确
程度直接影响到了编码的压缩性能。
-Motion estimation is the key to video encoding technology, its basic principle is the use of the adjacent time frame relevant, through the forecast period to reduce redundancy. In the actual code, in order to save bit rate and does not transfer all the data of each frame, but the use of motion estimation for each frame obtained with the prediction of the difference between the reference frame. Motion estimation more accurate, more closer to the margin and the distribution of zero, the smaller the energy difference between blocks, after transform, quantization and entropy coding of the code produced by the bit stream bit rate is less. Therefore, the motion estimation search accuracy of the direct impact on the performance of the compression coding.
Platform: |
Size: 17408 |
Author: 罗鹏 |
Hits:
Description: matlab编的基于图像矢量量化的水印程序，用分割码本的办法嵌入水印！-matlab for image vector quantization-based watermarking process, with the approach of the division of code embedded watermark!
Platform: |
Size: 2048 |
Author: 周赟 |
Hits:
Description: 用matlab实现基于LBG算法的向量量化编码-Matlab by LBG algorithm based on vector quantization coding
Platform: |
Size: 1024 |
Author: vvvpig |
Hits:
Description: The QccPackSPECK module is a optional addon to the QccPack library for
the quantization, compression and coding of data. -The QccPackSPECK module is a optional addon to the QccPack library for
the quantization, compression and coding of data. The QccPackSPECK module
provides an implementation of the Set-Partitioning Embedded Block (SPECK)
algorithm for image compression. Since the SPECK algorithm is patented,
it is made available here under special license terms (see below and the
file LICENSE-SPECK), whereas the QccPack library is licensed under the GNU
General Public License and the GNU Library General Public License. See the
file CHANGELOG-SPECK for details of recent changes to the QccPackSPECK code.
Platform: |
Size: 96256 |
Author: jason.. |
Hits:
Description: lvq vector quantization algorithm for the simulation program Matlab.
Platform: |
Size: 81920 |
Author: mony |
Hits:
Description: 产生长度为1000的标准正太分布的随机信号，画出时域波形及频谱；
采用u=255的非均匀PCM编码，每符号为8bit，画输入-输出关系图；
计算信号量化噪声比（SQNR）；
信道误码率为10-3；
解码，并画出u律反变换后的信号时域波形及频谱。
扩展要求：
采用均匀PCM、量化级数可变、信道误码率可变。
- have a standard length of 1000 is too random signal distribution, draw time-domain waveform and frequency spectrum u = 255 using the non-uniform PCM encoding, each symbol for 8bit, drawing input- output relationship diagram calculated quantization noise signal than (SQNR) channel bit error rate of 10-3 decoder, and law to draw u the signal after the inverse transform time-domain waveform and frequency spectrum. Expansion of the requirements: uniform PCM, quantization level variable, a variable channel bit error rate.
Platform: |
Size: 2048 |
Author: louzhu |
Hits:
Description: 将图像颜色量化为72种。可以做图像检索用的好东西-Images for 72 kinds of color quantization. Image retrieval can be done with good things! !
Platform: |
Size: 1024 |
Author: 孟 |
Hits:
Description: 直接频率合成DDS模块的ise工程，可以直接下载，在Spartan3/Spartan3E上验证通过。该DDS模块可以产生双通道的不同频率的正弦波，也可以产生同频的任意相位差的相移波形。本模块累加器位数为32位，可以产生12位相位精度12位量化精度的正弦波。该设计例化一个Block Ram，为节省储存空间仅需要储存1/4周期的数据。根据需要，可以重新修改数据，改变波形。-DDS direct frequency synthesizer module ，ise project, can be directly downloaded through the Spartan3/Spartan3E and tested successfully. The DDS module can generate two-channel sine wave of different frequency, or produce the same frequency arbitrary waveform phase difference of the phase shift. There is a 32-bit accumulator to generate 12 bit phase-precision 12-bit quantization precision of the sine wave. Cases the design of a Block Ram, in order to save storage space need to store only 1/4 cycle of data. Necessary, can modify data, change the waveform.
Platform: |
Size: 471040 |
Author: 郭先生 |
Hits:
Description: JPEG quantization for embedded systems
Platform: |
Size: 10240 |
Author: kingtut |
Hits:
Description: H264中的整数DCT变换编码及量化代码,并附有原文的DCT编码、量化的说明文档。本源码从X264及JM中抽取后整理得来，并用原文档中的示例数据进行了测试，与原文档结果完全一样，是学习H.264不可多得的参考资源.
-H264 in the integer DCT transform coding and quantization code, together with the original DCT coding, quantitative description of the document. The source from the X264 and JM Finishing extract obtained and used the original document in the test sample data with the results of exactly the same as the original document, is to study rare H.264 reference resources.
Platform: |
Size: 64512 |
Author: ren |
Hits:
Description: Wavelet Tree Quantization for
Copyright Protection Watermarking
Platform: |
Size: 855040 |
Author: mai |
Hits:
Description: 工程名为DctQuant，编程实现了一帧CIF格式4:2:0的YUV图像的H.264的整数变换、量化以及逆量化、逆变换。具体的功能表述如下：
1、首先按“打开YUV”按钮，选择一帧格式为CIF的4:2:0的YUV图像（默认的YUV图像的格式为4:2:0的CIF格式）。选择完成后，可以看到显示的输入图像。
2、输入QP值，QP是0到51之间的整数值，超出范围，会弹出对话框提示。QP默认为28。注意：当改变QP值为0时，即没有量化，则点击转换按钮显示出的PSNR值即为输入图像的PSNR值。
3、点击转换按钮，可以看到经过变换量化和逆量化逆变换过程以后输出的图像以及显示输出图像的PSNR值（保留小数点后6位）和默认4x4块号为0的原始各像素值、整数变换后的值、量化后、逆量化后、逆变换后的值。
4、可以手动改变块号，块号的范围为0-6335，超出范围会弹出对话框提示。然后单击“输入块号”按钮即可显示此块号对应的原始各像素值、整数变换后的值、量化后、逆量化后、逆变换后的值。
-Works entitled DctQuant, programming of a CIF format, 4:2:0 YUV images H.264 integer transform, quantization, and inverse quantization, inverse transform. Expression of specific functions are as follows:
1, first press the "Open YUV" button, select a format for the CIF of the 4:2:0 YUV image (the default YUV image format 4:2:0 CIF format). Selection is complete, you can see displayed in the input image.
2, enter the QP value, QP is an integer between 0-51 values out of range, will pop-up dialog box prompt. QP default is 28. Note: When changing the QP value is 0, that is not quantified, then click the convert button to show the PSNR values of the PSNR values shall be the input image.
3, click the Convert button, you can see through the transformation inverse quantization and inverse quantization output of the image after the transformation process and display the output image PSNR values (retention decimal 6), and the default 4x4 block No. 0 of the original values of each pixel , int
Platform: |
Size: 446464 |
Author: 王婉君 |
Hits:
Description: This a JPEG encoding & decoding program of still image. It does not use level shifting. Discrete Cosine transform (DCT) is performed both by classical & Chen s Flowgraph methods. Predefined JPEG quantization array & zigzag order are used here. RUN - LEVEL coding is used instead of Huffman coding. Compression ratio is compared for each DCT method. Effect of coarse and fine quantization is also examined. The execution time of 2 DCT methods is also checked. In addition, most energatic DCT coefficients are also applied to examine the effect in MatLab 7.4.0 R2007a. Input is 9 gray scale pictures & output is 9*9=81 pictures to compare. Blocking effect is obvious.
-This is a JPEG encoding & decoding program of still image. It does not use level shifting. Discrete Cosine transform (DCT) is performed both by classical & Chen s Flowgraph methods. Predefined JPEG quantization array & zigzag order are used here. RUN - LEVEL coding is used instead of Huffman coding. Compression ratio is compared for each DCT method. Effect of coarse and fine quantization is also examined. The execution time of 2 DCT methods is also checked. In addition, most energatic DCT coefficients are also applied to examine the effect in MatLab 7.4.0 R2007a. Input is 9 gray scale pictures & output is 9*9=81 pictures to compare. Blocking effect is obvious.
Platform: |
Size: 95232 |
Author: Ho Joon Lee |
Hits:
Description: Audio watermarking based on quantization in wavelet domain
Platform: |
Size: 2048 |
Author: Raj |
Hits:
«
1
2
3
4
5
67
8
9
10
11
...
50
»