公开(公告)号：US5864817A

公开(公告)日：1999-01-26

申请号：US548895

申请日：1995-10-26

申请人： David E. Galbi

发明人： David E. Galbi

IPC分类号： H04N5/922 ,  G06T9/00 ,  G10L19/00 ,  G10L19/02 ,  G11B20/10 ,  H03M7/00 ,  H03M7/42 ,  H04N5/21 ,  H04N5/44 ,  H04N5/60 ,  H04N5/92 ,  H04N7/26 ,  H04N7/30 ,  H04N7/50 ,  H04N7/52 ,  H04N21/2368 ,  H04N21/434 ,  G10L3/02

CPC分类号： H04N19/00 ,  G10L19/02 ,  H03M7/425 ,  H04N19/126 ,  H04N19/42 ,  H04N19/423 ,  H04N19/43 ,  H04N19/433 ,  H04N19/436 ,  H04N19/439 ,  H04N19/44 ,  H04N19/60 ,  H04N19/61 ,  H04N19/63 ,  H04N19/90 ,  H04N21/4341 ,  H04N5/4401 ,  H04N5/602 ,  G09G2340/02 ,  G10L19/005 ,  H04N19/124 ,  H04N19/13 ,  H04N19/70 ,  H04N19/91 ,  H04N5/21

摘要： An MPEG audio/video decoder has memories, a signal processing unit (SPU) including a multiplier and a butterfly unit, a main CPU, and a memory controller which are time division multiplexed between decoding video and audio data. For audio decoding, the butterfly unit determines combinations of components of a frequency-domain vector to reduce the number of multiplies required to transform to the time domain (matrixing). Matrixing is interwoven with MPEG filtering to increase throughput of the decoder by increasing parallel use of the multiplier, the butterfly unit, and a memory controller.

摘要翻译： MPEG音频/视频解码器具有存储器，包括在解码视频和音频数据之间进行时分复用的乘法器和蝶形单元，主CPU和存储器控制器的信号处理单元（SPU）。 对于音频解码，蝶形单元确定频域矢量的分量的组合以减少变换到时域所需的乘法数（矩阵）。 矩阵与MPEG滤波交织，通过增加乘法器，蝶形单元和存储器控制器的并行使用来提高解码器的吞吐量。

公开(公告)号：US5848195A

公开(公告)日：1998-12-08

申请号：US568021

申请日：1995-12-06

申请人： Joseph N. Romriell

发明人： Joseph N. Romriell

IPC分类号： G06T9/00 ,  H03M7/42 ,  H04N7/26 ,  H04N7/50 ,  H04N1/41

CPC分类号： H04N19/149 ,  H03M7/42 ,  H04N19/179 ,  H04N19/61 ,  H04N19/619 ,  H04N19/63 ,  H04N19/10 ,  H04N19/13 ,  H04N19/91

摘要： Selecting a Huffman table to encode a set of signals, such as video signals, from a set of predefined Huffman tables. A histogram is generated for the set of signals and used to determine the number of bits to encode the set of signals for each of the predefined Huffman tables. The Huffman table that provides the smallest number of bits to encode is selected for encoding the set of signals into the encoded bitstream. In a preferred embodiment, a dynamic Huffman table is also generated for the set of signals and used to estimate the number of bits to encode the set of signals including the number of bits to encode the dynamic Huffman table. If using the dynamic Huffman table results in fewer bits in the encoded bitstream, then the dynamic Huffman table is used (and explicitly encoded in the bitstream) instead of the selected predefined Huffman table.

摘要翻译： 选择霍夫曼表从一组预定义霍夫曼表中编码一组信号，如视频信号。 为该组信号生成直方图，并用于确定为每个预定霍夫曼表格编码信号组的位数。 选择提供要编码的最小位数的霍夫曼表，以将该组信号编码为编码比特流。 在优选实施例中，还为该组信号生成动态霍夫曼表，并且用于估计包括要编码动态霍夫曼表的比特数的信号集合的比特数。 如果使用动态霍夫曼表导致编码比特流中的较少位，则使用动态霍夫曼表（并且在比特流中显式地编码），而不是所选择的预定霍夫曼表。

公开(公告)号：US5831560A

公开(公告)日：1998-11-03

申请号：US718201

申请日：1996-09-20

申请人： Alexander D. Gelfenbain

发明人： Alexander D. Gelfenbain

IPC分类号： H03M7/42 ,  H03M7/38

CPC分类号： H03M7/42

摘要： The present invention is a method and apparatus for translating an input character from a source codeset to a target codeset. Structurally, the present invention includes a list of translation tables each having one or more entries. Each entry in a translation table defines a translation from a character in the source codeset to a character in the target codeset. Each translation table also includes an entry condition function and an input normalization function. To translate an input character in the source codeset to the target codeset, the list of translation tables is traversed until an entry condition is found which returns a boolean value of true when called with the input character is passed as a parameter. Then input character is then passed to the input normalization function which returns the index or position of the translation for the input character in the translation table. The contents of the translation table, at the index, is then returned as the translation of the input character.

摘要翻译： 本发明是用于将输入字符从源代码集转换为目标代码集的方法和装置。 在结构上，本发明包括每个具有一个或多个条目的翻译表的列表。 翻译表中的每个条目定义从源代码集中的字符到目标代码集中的字符的转换。 每个转换表还包括输入条件函数和输入归一化函数。 要将源代码集中的输入字符转换为目标代码集，将遍历转换表的列表，直到发现输入条件，当输入字符作为参数传递时，返回布尔值为true的值。 然后，输入字符然后传递给输入归一化函数，该函数返回翻译表中输入字符的翻译的索引或位置。 然后，在索引处的翻译表的内容作为输入字符的翻译返回。

公开(公告)号：US5825312A

公开(公告)日：1998-10-20

申请号：US758137

申请日：1996-11-25

申请人： Remo J. D'Ortenzio

发明人： Remo J. D'Ortenzio

IPC分类号： H03M7/42 ,  H03M7/40

CPC分类号： H03M7/425

摘要： A low cost, high speed, JPEG Huffman code decoder. The entire gamut of Huffman codes is partitioned into groups, with each group being associated with it's own small look up table to minimize the overall memory requirements. The current Huffman code is stored in a register. For the disclosed embodiment there are N=4 memories with sizes 128, 256, 128 and 256 elements respectively. For the specific typical Huffman code set cited, the partitioning is: The code words in the first group have 7 bits or less and are decoded in the first memory. The next group has from 5 to 7 leading ones, where the first five 1's are masked out and the remaining bits are used to address the second memory. The remaining codes have at least eight leading ones. Here, the first eight ones are masked out and the remaining bits are used to address the last memory. (In this example, one of the memories is not used). Finally, the first eight bits are used in a selector to enable the correct memory so that a single correct output data word is output. The disclosed N=4 embodiment is designed to handle any custom JPEG Huffman code set. This concept can be extended to other memory partitions (a N=6 example is cited) that might provide some desirable memory vs logic switching tradeoffs.

摘要翻译： 低成本，高速度的JPEG霍夫曼码解码器。 Huffman代码的整个区域被分成几组，每个组都与自己的小型查找表相关联，以最大限度地减少整体内存需求。 当前的霍夫曼代码存储在一个寄存器中。 对于所公开的实施例，分别具有大小128,256,128和256个元素的N = 4个存储器。 对于所引用的具体典型的霍夫曼代码集，划分是：第一组中的码字具有7位或更少，并且在第一存储器中被解码。 下一组有5到7个，其中前5个被屏蔽，其余的位用于寻址第二个存储器。 剩余的代码至少有八个主要代码。 这里，前八个被屏蔽，其余的位用于寻址最后的存储器。 （在该示例中，不使用其中一个存储器）。 最后，前8位用于选择器以启用正确的存储器，以便输出单个正确的输出数据字。 所公开的N = 4实施例被设计为处理任何定制的JPEG霍夫曼代码集。 这个概念可以扩展到可能提供一些期望的存储器与逻辑切换权衡的其他存储器分区（N = 6的例子）。

公开(公告)号：US5822770A

公开(公告)日：1998-10-13

申请号：US581254

申请日：1995-12-29

申请人： Dae-yun Shim

发明人： Dae-yun Shim

IPC分类号： H04N19/423 ,  H03M7/40 ,  H03M7/42 ,  H04N1/41 ,  H04N7/24 ,  H04N19/00 ,  H04N19/44 ,  H04N19/91 ,  G06F13/00

CPC分类号： H03M7/425

摘要： A memory interfacing method for a variable-length decoder and a circuit therefor are provided. The memory interfacing circuit includes a read controller for generating a read chip-enable signal and a read address in response to a first request signal output from the variable-length decoder, generating a second request signal when the read address reaches a predetermined level, and providing the second request signal to an external memory controller which is connected to an external memory, a write controller for generating a write address and a write section signal in response to an accept signal generated by the external memory controller in response to the second request signal, and an internal memory controller which includes an internal memory, for controlling data provided from an external memory to be written according to the write address and the write section signal generated by the write controller. Accordingly, the present invention simplifies complex signal processing due to frequent memory access between a variable-length decoder and an external memory.

摘要翻译： 提供了一种用于可变长度解码器及其电路的存储器接口方法。 存储器接口电路包括：读控制器，用于响应于从可变长度解码器输出的第一请求信号产生读取芯片使能信号和读取地址，当读取地址达到预定电平时产生第二请求信号;以及 向连接到外部存储器的外部存储器控制器提供第二请求信号，响应于外部存储器控制器响应于第二请求信号而产生的接受信号而产生写入地址和写入部分信号的写入控制器 以及内部存储器控制器，其包括内部存储器，用于根据由写入控制器产生的写入地址和写入部分信号控制从外部存储器提供的要写入的数据。 因此，本发明通过可变长度解码器和外部存储器之间的频繁存储器访问来简化复杂的信号处理。

公开(公告)号：US5801650A

公开(公告)日：1998-09-01

申请号：US566583

申请日：1995-11-28

申请人： Tadayoshi Nakayama

发明人： Tadayoshi Nakayama

IPC分类号： H04N19/00 ,  H03M7/40 ,  H03M7/42 ,  H04N1/41 ,  H04N19/426 ,  H04N19/44 ,  H04N19/625 ,  H04N19/91

CPC分类号： H03M7/425

摘要： An unpacking circuit 103 takes the top 16-bit data from an input code after eliminating previously decoded data on the basis of code length information 106. An operation unit 501 obtains a redundant code number of a Huffman code inputted from the unpacking circuit 103 and upper three bits of the code length information 106. By referring to a decoding table 503, an eight-bit fixed-length data and the least significant bit of the code length information are obtained correspondence with an input nine-bit redundant code number signal 502. Therefore, memory size required for storing the decoding table 503 is 2.sup.9 .times.9 bits=576 bytes, thus, comparing to a case where memory size for storing the decoding table of 96 kbytes when changing into a fixed code length by using only the decoding table, it is possible to use a decoding table of smaller size.

摘要翻译： 拆包电路103在根据码长度信息106消除先前解码的数据之后，从输入代码中取出前16位数据。操作单元501获得从拆包电路103输入的霍夫曼代码的冗余码号和上部 通过参考解码表503，与输入的九位冗余码数信号502相对应地获得八位固定长度数据和码长度信息的最低有效位。 因此，存储解码表503所需的存储器大小为29×9比特= 576字节，因此，与仅使用解码表变更为固定码长的情况相比，存储96K字节的解码表的存储器大小的情况 可以使用较小尺寸的解码表。

公开(公告)号：US5787210A

公开(公告)日：1998-07-28

申请号：US532035

申请日：1995-09-22

申请人： Sang-Ho Kim

发明人： Sang-Ho Kim

IPC分类号： H04N19/00 ,  G06T5/20 ,  G06T9/00 ,  H03M7/42 ,  H04N1/41 ,  H04N5/21 ,  H04N19/117 ,  H04N19/167 ,  H04N19/182 ,  H04N19/192 ,  H04N19/196 ,  H04N19/423 ,  H04N19/625 ,  H04N19/80 ,  H04N19/86 ,  H04N19/91 ,  G06K9/03 ,  G06K9/00 ,  H04N11/04 ,  H04N17/02

CPC分类号： H04N19/86 ,  H04N5/21

摘要： A post-processing method for filtering decoded image signals determines positions of target pixels to be filtered and generates a selection signal indicating whether or not each of the target pixels belongs to a boundary region of a block of decoded image data. In response to the selection signal, each of the target pixel data is selectively filtered through the use of a filtering unit having two filters with different cutoff frequencies. An absolute difference value between filtered and non-filtered target pixel values is compared with a predetermined threshold value. The target pixel value is updated with the filtered target pixel value if the absolute difference value is smaller than the predetermined threshold value; otherwise, a compensated target pixel value is obtained by subtracting the predetermined threshold value from the non-filtered target pixel value or adding the non-filtered target pixel value and the predetermined threshold value, and, then the filtered target pixel value is updated with the compensated target pixel value.

摘要翻译： 用于对解码图像信号进行滤波的后处理方法确定要滤波的目标像素的位置，并产生指示每个目标像素是否属于解码图像数据块的边界区域的选择信号。 响应于选择信号，通过使用具有不同截止频率的两个滤波器的滤波单元来选择性地滤波每个目标像素数据。 滤波和未滤波的目标像素值之间的绝对差值与预定的阈值进行比较。 如果绝对差值小于预定阈值，则用滤波的目标像素值更新目标像素值; 否则，通过从未滤波的目标像素值中减去预定阈值或将未滤波的目标像素值和预定阈值相加来获得补偿目标像素值，然后用 补偿目标像素值。

公开(公告)号：US5751233A

公开(公告)日：1998-05-12

申请号：US644246

申请日：1996-05-10

申请人： Tetsuya Tateno ,  Yuji Minami

发明人： Tetsuya Tateno ,  Yuji Minami

IPC分类号： G06T9/00 ,  H03M7/42 ,  H03M7/44 ,  H04N1/419 ,  H03M7/40

CPC分类号： G06T9/005 ,  H03M7/425 ,  H04N1/419 ,  H04N19/13 ,  H04N19/91

摘要： A decoder decodes input codes, such as Modified Huffman, Modified READ, and Modified Modified READ codes, and includes a zero bit detector which detects the number of consecutive leading zero bits of the input code. An address compressor forms address data by performing a logical operation of data indicating the number of detected zero bits and data excluding the consecutive leading zero bits and the next one bit of the data. A reference table for code conversion is addressed by the formed address data from the address compressor and outputs decoded data corresponding to the input code.

摘要翻译： 解码器对诸如Modified Huffman，Modified READ和Modified Modified READ码的输入代码进行解码，并且包括检测输入代码的连续前导零位数的零位检测器。 地址压缩器通过执行表示检测到的零比特数的数据和除了连续前导零比特和数据的下一个比特之外的数据的逻辑运算来形成地址数据。 通过来自地址压缩器的形成的地址数据来解决用于代码转换的参考表，并且输出与输入代码对应的解码数据。

公开(公告)号：US5748121A

公开(公告)日：1998-05-05

申请号：US568247

申请日：1995-12-06

申请人： Joseph N. Romriell

发明人： Joseph N. Romriell

IPC分类号： G06T9/00 ,  H03M7/42 ,  H04N7/26 ,  H03M7/40 ,  H03M7/30

CPC分类号： H03M7/42 ,  H04N19/1883 ,  H04N19/61 ,  H04N19/63 ,  H04N19/70 ,  H04N19/13 ,  H04N19/91

摘要： Generating Huffman tables based on off-line analysis of training sets of signals. Individual Huffman tables are generated for each training set, preferably from an individual histogram representing the training set. The training sets are grouped together, starting with a single group containing all of the training sets. A group Huffman table is generated for each group. A group is selected for splitting and a new Huffman table is generated for a new group. The group to be split is selected by identifying the group that is most inefficiently encoded using a single Huffman table. The new Huffman table is generated by perturbing the group histogram for the group to be split and then generating the new Huffman table from the perturbed histogram. After defining the new Huffman table for the new group, the training sets are redistributed among the existing groups. A training set is moved to a different group if the different group's Huffman table does a better job of encoding that training set than the Huffman table for the training set's old group. After the training sets are redistributed, the group Huffman tables are regenerated. Training-set redistribution is repeated until the redistribution processing converges (e.g., no training sets change groups). Groups are continued to be split until the desired number of group Huffman tables have been generated.

摘要翻译： 基于训练集信号的离线分析生成霍夫曼表。 针对每个训练集生成单个霍夫曼表，优选地从表示训练集的单个直方图。 训练集合在一起，从包含所有训练集的单个组开始。 为每个组生成一个组霍夫曼表。 选择一个组进行拆分，并为新组生成一个新的霍夫曼表。 通过使用单个霍夫曼表来识别最无效编码的组来选择要拆分的组。 通过扰乱要分割的组的组直方图，然后从干扰的直方图中生成新的霍夫曼表，生成新的霍夫曼表。 在为新组定义新的霍夫曼表之后，训练集在现有组中重新分配。 如果不同组的霍夫曼表比训练集的旧组的霍夫曼表更好地编码训练集，训练集被移动到不同的组。 在重新分配训练集之后，重新生成霍夫曼组。 重复训练集再分配直到再分配处理收敛（例如，没有训练集改变组）。 组继续分割，直到生成所需数量的组霍夫曼表。

公开(公告)号：US5745504A

公开(公告)日：1998-04-28

申请号：US668460

申请日：1996-06-25

申请人： Karl Goran B.ang.ng

发明人： Karl Goran B.ang.ng

IPC分类号： H03M7/42 ,  H03M13/00 ,  H04L1/00

CPC分类号： H04L1/0059 ,  H03M7/42 ,  H04L1/0071

摘要： Data is transmitted and received over a channel in a telecommunications system includes encoding the data by providing a plurality of fixed length code tables, wherein each of the fixed length code tables has a size that is different from a size of every other fixed length code table, and encoding the data as a plurality of information code words, wherein each information code word is a selected value from one of the fixed length code tables. A plurality of marker codes is also provided, wherein the marker codes are variable length codes. A unique one of the marker codes is then associated with each of the fixed length code tables, and channel code words are generated by inserting the associated marker code one bit at a time into each of the information code words at a fixed bit interval. The encoded data may be decoded by processing in a forward direction to extract and decode marker code words. Information code bits associated with valid decoded marker code words are extracted and decoded. When an invalid marker code word is extracted, backwards processing is commenced to extract and decode marker code words until an invalid marker code is detected in the backwards direction. An error interval is defined by the positions of the invalid marker code words detected in the forward and backwards directions. Only those valid information code words associated with marker code words that lie outside of the error interval are output for further processing.

摘要翻译： 通过电信系统中的信道发送和接收数据包括通过提供多个固定长度码表来对数据进行编码，其中固定长度码表中的每一个具有与每隔一个固定长度码表的大小不同的大小 并且将数据编码为多个信息码字，其中每个信息码字是来自固定长度码表之一的选择值。 还提供了多个标记代码，其中标记代码是可变长度代码。 然后将唯一的一个标记代码与每个固定长度码表相关联，并且通过以固定的位间隔一次将相关联的标记码一位地插入每个信息码字中来生成信道码字。 编码数据可以通过正向处理来解码，以提取和解码标记码字。 提取和解码与有效解码标记码字相关联的信息码位。 当提取无效的标记代码字时，开始向后处理以提取和解码标记代码字，直到在向后方向上检测到无效的标记代码。 错误间隔由在向前和向后方向上检测到的无效标记代码字的位置定义。 仅输出与错误间隔之外的标记代码字相关联的有效信息代码字，用于进一步处理。