CN103841421A - USDC data processing method - Google Patents
USDC data processing method Download PDFInfo
- Publication number
- CN103841421A CN103841421A CN201410077400.2A CN201410077400A CN103841421A CN 103841421 A CN103841421 A CN 103841421A CN 201410077400 A CN201410077400 A CN 201410077400A CN 103841421 A CN103841421 A CN 103841421A
- Authority
- CN
- China
- Prior art keywords
- image
- usdc
- data processing
- processing method
- coding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a USDC data processing method which comprises the following steps of color image wavelet decomposition, wherein hypercomplex wavelet transform is adopted, and color information of color images is maintained; vector encoding, wherein a hypercomplex cluster analysis process is adopted, and human eye sensitive information content in the images is maintained; motion compensation, wherein asymmetric crossed multilevel hexagonal-grid improved search is adopted, classification is carried out according to motion intensity, and compressed bit flow is obtained; finally the compressed bit flow is output through a network. The technological mode of combination of the color image wavelet decomposition, the vector encoding and the motion compensation is adopted, so that video transmission code flow becomes small, video images are compressed according to the image conditions, bandwidth is greatly saved, delay is reduced, and operating cost is saved.
Description
Technical field
The present invention relates to computer realm encoding video pictures field, relate in particular to a kind of USDC data processing method.
Background technology
At present, the actual video encoding standard in the world is the H.264/MPEG-4AVC standard of the U.S., this is a kind of block-oriented video compression standard that damages based on motion compensation, H.264 be by ITU-T (ITU Telecommunication Standardization Sector, International Telecommunications Union's telecommunication standards group) the high compression digital video coding-coding device standard that proposes of the joint video team (JVT, Joint Video Team) that constitutes jointly of Video coding expert group (VCEG) and ISO/IEC dynamic image expert group (MPEG).H.264 one of standard that is ITU-T take series H.26x as name nominating, AVC is ISO/IEC MPEG mono-side's address simultaneously.It is to pass through in the 7th meeting of JVT of holding at Thailand Pattaya in March, 2003 that this standard formally becomes international standard.Because this standard is to formulate by two different tissues are common, therefore there are two different titles: in ITU-T, it be named as H.264; And in ISO/IEC, it is called as the 10th part of MPEG-4, i.e. advanced video coding (AVC).
H.264 on coding framework, still continue to use MC-DCT structure in the past, it is mixing (hybrid) structure that motion compensation adds transition coding, therefore it has retained the feature of some previous standards, as unrestricted motion vector (unrestricted motion vectors), the median prediction (median prediction) to motion vector etc.But, be H.264 greatly improved in performance than video encoding standard before.This improves does not depend merely on a certain technology and realizes, but the little improvement in performance of being brought by various different technologies and common generation.This wherein main technology have:
1. infra-frame prediction
To the coding of I frame by utilizing spatial coherence but not temporal correlation realize.Standard in the past has only been utilized the inner correlation of a macro block (macroblock), and has ignored the correlation between macro block, so the data volume after general coding is larger.In order further to utilize spatial coherence, H.264 introduce infra-frame prediction to improve compression efficiency.Briefly, intraframe predictive coding is exactly to predict current pixel value with contiguous pixel value around, then predicated error is encoded.This prediction is block-based, and for luminance component (luma), the size of piece can be selected between 16 × 16 and 4 × 4, and 16 × 16 have 4 kinds of predictive modes, and 4 × 4 have 9 kinds of predictive modes; For chromatic component (chroma), prediction is carried out whole 8 × 8, has 4 kinds of predictive modes.Except DC prediction, the prediction on other the every kind corresponding different directions of predictive mode.
2. inter prediction
The same with standard in the past, H.264 eliminate time redundancy with Motion estimation and compensation, but it has following five different features:
(1) variable size of used while prediction
Because all pixels in block-based motion model hypothesis piece have all been done identical translation, can be quite different with reality in motion this hypothesis of edge relatively more violent or moving object, thereby cause larger predicated error, the size that at this moment reduces piece can make hypothesis still set up in little piece.The blocking effect that in addition little piece causes is relatively also little, so in general little piece can improve the effect of prediction.For this reason, H.264 adopted altogether 7 kinds of modes to cut apart a macro block, under every kind of mode, the size and shape of piece is not identical, and this just makes the predictive mode that encoder can be best according to the content choice of image.With only use 16 × 16 predict compared with, use the piece of different size and shapes can make code check save more than 15%.
(2) meticulousr precision of prediction
In H.264, the motion vector of Luma component (MV) uses 1/4 pixel precision.The MV of Chroma component is derived by luma MV, because chroma resolution is the half (to 4: 2: 0) of luma, so its MV precision will be 1/8, this that is to say that the displacement of the chroma MV representative of 1 unit is only 1/8 of chroma component sampling point spacing.So meticulous precision of prediction can make code check saving exceed 20% than integer precision.
(3) multi-reference frame
H.264 support multiple reference frame prediction (multiple reference frames), can have the frame of decoding before present frame of more than one (maximum 5) to can be used as reference frame and produce the prediction (motion-compensated prediction) to present frame.This is applicable to the situation that contains cycle movement in video sequence.Adopt this technology, can improve the performance of estimation (ME), improve the H.264 error recovery capabilities of decoder, but also increased the complexity of capacity and the codec of buffer memory simultaneously.But, proposition is H.264 the develop rapidly of based semiconductor technology, and therefore these two burdens can become insignificant in the near future.Than only using a reference frame, use 5 reference frames can save code check 5～10%.
(4) anti-block effect filtering device
Anti-block effect filtering device (Deblocking Filter), its effect is to eliminate after inverse quantization and inverse transformation, to rebuild the blocking effect producing due to predicated error in image, be the pixel value saltus step at block edge place, thereby one improve the subjective quality of image, two reduce predicated error.H.264 the Deblocking Filter in can also judge according to picture material, only the pixel value saltus step producing due to blocking effect is carried out smoothly, and is retained the pixel value of objects in images edge is discontinuous, in order to avoid cause edge blurry.Different from Deblocking Filter is in the past, image after filtering will be placed in buffer memory as required for inter prediction, rather than only rebuild and be used for improving subjective quality when image in output, that is to say that this filter is arranged in decoding ring but not outside the output of decoding ring, thereby it is called again Loop Filter.It should be noted that for infra-frame prediction, use be the reconstruction image of filtered ripple.
3. integer transform
H.264 in frame or the residual error of inter prediction (residual) carry out dct transform coding.The hardware designs complexity of bringing in order to overcome floating-point operation, the more important thing is the problem of not mating (mismatch) between the encoder that rounding error causes, new standard is revised the definition of DCT, conversion only can be realized with integer addition and subtraction and shifting function, in the situation that not considering quantization influence, the output of decoding end can recover the input of coding side exactly like this.Certainly the cost of doing is like this decline slightly of compression performance.In addition, this conversion is carried out for 4 × 4, and this also contributes to reduce blocking effect.In order further to utilize the spatial coherence of image, after the prediction residual of the prediction residual to chroma and 16 × 16 infra-frame predictions is carried out above-mentioned Integer DCT Transform, standard also, by the piece of DC coefficient composition 2 × 2 or 4 × 4 sizes in each 4 × 4 transformation coefficient blocks, is further done Hadamard (Hadamard) conversion.
4. entropy coding
If Slice layer prediction residual, H.264 there is the mode of two kinds of entropy codings: based on contextual adaptive variable length code (Context-based Adaptive Variable Length Coding, CAVLC) with based on contextual adaptive binary arithmetic coding (Context-based Adaptive Binary Arithmetic Coding, CABAC); If not prediction residual, H.264 adopt Exp-Golomb code or CABAC coding, depending on the setting of encoder.
(1)CAVLC
The basic thought of VLC is exactly that the large symbol of the frequency of occurrences is used to shorter code word, and the little symbol of the frequency of occurrences adopts longer code word.Can make like this mean code length minimum.In CAVLC, H.264 adopt some VLC code tables, different probabilistic model corresponding to code table.Encoder can be based on context, as the around nonzero coefficient of piece or the order of magnitude of coefficient, in these code tables, automatically selects, and mates most possibly, thereby realized the function of context-adaptive with the probabilistic model of current data.
(2)CABAC
Arithmetic coding is a kind of entropy encoding scheme efficiently, and the corresponding code length of its each symbol is considered to mark.Due to relevant with the result of former coding to the coding of each symbol, be the probability nature of source symbol sequence entirety so its considers, rather than the probability nature of single symbol, thereby it can approach the limit entropy of information source to a greater degree, reduces code check.In order to get around the problem of representation of infinite precision decimal in arithmetic coding and source symbol probability to be estimated, modern arithmetic coding is realized mainly with the mode of finite state machine, and CABAC is H.264 an example, and other example also has JPEG2000.In CABAC, binary character of every coding, encoder will adjust automatically to the estimation of (representing with " state ") of information source probabilistic model, on the probabilistic model basis that binary character subsequently has just upgraded at this, encodes.Such encoder does not need the priori of information source statistical property, but estimates adaptively in cataloged procedure.Obviously, with the method for pre-set some probabilistic models in CAVLC coding by comparison, CABAC has greater flexibility, can obtain better coding efficiency---the reduction of about 10% code check.The feature of more than introducing is all for improving coding efficiency H.264, in addition H.264 also there is good error recovery capabilities (error resilience) and network adaptability (network adaptability), introduce Some features wherein below.
5.SP?Slice
The main purpose of SP Slice is the switching (switch) for different code streams, also can be used in addition random access, fast forwarding and fast rewinding and wrong recovery of code stream.Here said different code streams refers under different bit rates restriction same information source the produced code stream of encoding.If the last frame before switching in transmission code stream is Al, target code stream the first frame after switching is B2 (supposing it is P frame), and because the reference frame of B2 does not exist, so directly switching obviously can cause very large distortion, and this distortion meeting is transmitted backward.Simple solution is exactly a B2 for transmission intraframe coding, but the data volume of general I frame is very large, and this method can cause the suddenly increase of transmission code rate.According to hypothesis above, because being encodes to same information source, although bit rate difference, but two frames before and after switching must have a lot of correlations, so reference frame that encoder can be using A1 as B2, B2 is carried out to inter prediction, and predicated error is exactly SP Slice, then completes the switching of code stream by transmitting SP Slice.Different from conventional P frame, the prediction that generation SP Slice carries out is to carry out in the transform domain of A1 and B2.It is the same when after SPSlice requires to switch, the image of B2 should be with direct transmission target code stream.Obviously,, if the target of switching is another code stream of wide of the mark, SP Slice is just inapplicable.
6. macro block sequence flexibly
(the flexible macroblock ordering of macro block sequence flexibly, FMO), refer to the macro block in piece image is divided into several groups, absolute coding respectively, macro block in some group not necessarily under conventional scanning sequency before and after continuously, and may be each diverse location being dispersed in randomly in image.If time made a mistake in transmission like this, when some macro block in certain group can not be correctly decoded, decoder still can according to the spatial coherence of image rely on its around the pixel of correct decoding it is recovered.
Therefore, prior art has yet to be improved and developed.
Summary of the invention
In view of above-mentioned the deficiencies in the prior art, the object of the present invention is to provide a kind of USDC data processing method, compress according to image situation video image, save greatly bandwidth, reduce time delay, save operation cost.
Technical scheme of the present invention is as follows:
A kind of USDC data processing method, it comprises the following steps:
Comprise coloured image wavelet decomposition step, adopt supercomplex wavelet transformation, input picture is decomposed into the subgraph of different frequency, wherein low frequency subgraph similarly is the approximate of original image, the high frequency subgraph of three directions has reflected the detailed information of original image, retains the colouring information of coloured image;
Comprise vector coding step, adopt hypercomplex process of cluster analysis, retain human eye sensitive information content in image;
Comprise motion compensation step, adopt asymmetric cross multi-level hexagon lattice point to improve search, classify according to exercise intensity; Use the expansion of hexagon template simultaneously, the key character information of the temporal correlation between frame of video and image self is compressed, obtain the bit stream of compression;
Finally the bit stream of above-mentioned compression is exported from network.
Described USDC data processing method, wherein, adopts macroblock coding unit to carry out coloured image wavelet decomposition step.
Described USDC data processing method, wherein, adopts motion estimation unit to carry out vector coding step.
Described USDC data processing method, wherein, adopts restructure from motion unit to carry out motion compensation step.
A kind of USDC data processing method provided by the invention, the technological means that adopts coloured image wavelet decomposition step, vector coding step to combine with motion compensation step, transmission of video code stream is diminished, compared with technology H.264, under identical image quality, code stream only has its 1/4 to 1/10 size, has saved greatly bandwidth, and transmission of video time delay is little, general video time delay is less than 1 second, is almost equivalent to real-time video; Adopt the inventive method aspect early investment, can save the equipment investment such as digital camera, DVR; Aspect operation input, transmission of video data traffic can be saved more than 75%, and video storage space can be saved more than 80%, has significantly saved operation cost; Traditional H.264 technology monitor video signal to noise ratio is about 38db left and right, and the monitor video signal to noise ratio of USDC of the present invention coding can reach 43db, has increased substantially video definition.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of USDC data processing method in the present invention;
Fig. 2 is the structural representation of ping-pong buffer structure in the present invention;
Fig. 3 adopts the present invention to carry out the idiographic flow schematic diagram of data processing.
Embodiment
The invention provides a kind of USDC data processing method, for making object of the present invention, technical scheme and effect clearer, clear and definite, below the present invention is described in more detail.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.
The invention provides a kind of USDC data processing method, as shown in Figure 1, it comprises the following steps:
Comprise coloured image wavelet decomposition step, adopt supercomplex wavelet transformation, input picture is decomposed into the subgraph of different frequency, wherein low frequency subgraph similarly is the approximate of original image, the high frequency subgraph of three directions has reflected the detailed information of original image, retains the colouring information of coloured image;
Comprise vector coding step, adopt hypercomplex process of cluster analysis, retain human eye sensitive information content in image;
Comprise motion compensation step, adopt asymmetric cross multi-level hexagon lattice point to improve search, classify according to exercise intensity; Use the expansion of hexagon template simultaneously, the key character information of the temporal correlation between frame of video and image self is compressed, obtain the bit stream of compression;
Finally the bit stream of above-mentioned compression is exported from network.
Further, adopt macroblock coding unit to carry out coloured image wavelet decomposition step; Adopt motion estimation unit to carry out vector coding step; Adopt restructure from motion unit to carry out motion compensation step.
In order further to describe the present invention, below enumerate more detailed explanation.
The basic principle of USDC coding techniques is take image wavelet transform and classified vector quantization and searching algorithm as core.Traditional Wavelet Transformation Algorithm is widely used already in signal process field.In various compression algorithms, the quantification to wavelet coefficient and coding are the focus that people study all the time.Because the spatial distribution of wavelet coefficient has the tissue characteristic of multiresolution, make the coefficient of wavelet decomposition of image in locus and content, all have certain correlation, this feature makes these coefficients be well suited for adopting vector quantization method processing.Vector quantization technology not only can be avoided the redundancy encoding brought due to the loose distribution in image wavelet decomposition coefficient space and too high complexity, but also can describe preferably the correlation between these coefficients.
The formation of code book is one of core technology of vector quantization, is organized in certain sequence vector to be encoded by multiple wavelet coefficients, is called code book.Codebook Design is also appropriate another important content quantizing, i.e. the sorted target vector set of code book, and its vector is called code word.Code book is called as classification and searching method to the sorting algorithm of code word.Distortion criterion is the important indicator of evaluation algorithms.Good sorting technique not only wants distortion little, and wants speed fast.Current classification vector coding method mainly adopts LBG algorithm and various improvement algorithm, also has the neural network classification method of SOFM etc.Although these method principles are simple, algorithm structure complexity, computing length consuming time.Because mostly depend on inceptive code book design, its real-time and practicality have been affected simultaneously.For the problems referred to above, USDC coding techniques improves, method after improvement is under compared with the prerequisite of small loss classifying quality, accelerate the search speed of code book, simultaneously, the method, under the prerequisite of given distortion measure, makes the code book quantity of code book change with the variation of picture material, can adjust adaptively classification results.After classification, this algorithm is organized by different sub-band and in advance specific ordinal ranking code book, make it not only be applicable to rear class RLC (distance of swimming) coding, also be beneficial to the realization of implements spatial scalable compression, simultaneously, USDC coding techniques classified index value difference divides rear employing Huffman coding, thereby a step has improved algorithm performance.Concrete algorithm flow chart as shown in Figure 3.
The place that USDC coding techniques is different from general wavelet decomposition is, adopt the special wavelet transformation technique for coloured image, coloured image is converted to special vector space to be represented, design vector wavelet decomposition algorithm, thereby directly coloured image is processed, avoid colour loss, improved picture quality.The key step of related algorithm be summarised as following some:
1) wavelet decomposition of coloured image
Adopt supercomplex to represent coloured image, retained the colouring information of coloured image.On this basis, studied hypercomplex wavelet transformation.Tower traditional Mallat decomposition algorithm is incorporated into supercomplex field, input picture is decomposed into the subgraph of different frequency, wherein low frequency subgraph similarly is the approximate of original image, and the high frequency subgraph of three directions has reflected the detailed information of original image.Image is after multilevel wavelet decomposes, and its sub-band coefficients has following feature:
(a) retained the colouring information of image.
(b) the main energy information of image concentrates on the subband (LLk) of low frequency, and this sub-band division yardstick is thick, and frequency is low, higher to the sensitivity of picture structure.
(c) coefficient table of HL, LH, HH subband reveals obvious direction characteristic.
(d) between unidirectional each sub-band coefficients, there is stronger correlation.
2) vector coding
The process of cluster analysis of vector is called vector coding.What use was more at present is LBG and improves algorithm etc., although these algorithm explicit physical meaning, but shortcoming is also very outstanding: encoding efficiency depends on inceptive code book and exist simultaneously the problem of the large and length consuming time of amount of calculation, in addition, the number of symbols of code book needs to determine in advance, can not for concrete image adaptive regulate.For this reason, USDC coding techniques is guaranteeing under the prerequisite that encoding efficiency is less, to have adopted one based on hypercomplex vector encoding method on affect, has retained in image human eye sensitive information content more, has overcome to a certain extent the deficiency of conventional method.
3) motion compensation
Image sequence is the basic comprising of video image, has stronger correlation in sequence between each two field picture.The final purpose of image compression is exactly to reduce as far as possible the various correlations of image information, redundant information is reduced as far as possible, also need to consider human visual system's characteristic (Human Visual System, HVS) simultaneously, thereby reach higher compression ratio.For single-frame images, the available the whole bag of tricks introduced above reduces spatial coherence, and reduces image correlation structurally by methods such as fractal images; And for sequence of video images, improve compression ratio thereby should make full use of its correlation in time.Under normal circumstances, between consecutive frame image, there will be situation about differing greatly, thereby motion compensation technique is one of necessary links reducing frame-to-frame correlation.
The search of motion vector is one of difficult point of motion compensation.Owing to may having local structural change between consecutive frame image, so conventionally image is divided into multiple sub-blocks, each sub-block is searched for respectively to its motion vector, this method matching degree is higher, but algorithm complexity, link the most consuming time in whole cataloged procedure, accounts for greatly 60% of the whole scramble time, and the method for solution is to find the fast search algorithm of motion vector.USDC coding techniques has adopted asymmetric cross multi-level hexagon lattice point to improve searching algorithm (Improved UMHexagonS), classify according to exercise intensity, reduce to cut apart the selection of pattern, increase the anticipation algorithm of initial search point, improved the efficiency of retrieval motion vector; Use the expansion of hexagon template simultaneously, improved the precision of search, the method has very fast arithmetic speed and effect preferably, and because algorithm has been considered temporal correlation between frame of video and the key character information of image self, robustness is stronger.
In sum, the feature that the present invention has can be summarized as follows:
(a) employing of coloured image Wavelet Transformation Algorithm, has improved picture quality;
(b) vector classification completes in real time, is conducive to hardware and realizes;
(c) reduce code book search arithmetic amount, improved algorithm speed;
(d) vector assorting process is with motion compensation with mate simultaneously and complete;
(e) improved the motion estimation vectors search speed of improved UMHexagonS algorithm;
USDC coding techniques is realized Real-time Video Compression, has first improved Cache hit rate by following three kinds of methods:
1), whole encryption algorithm is divided into 3 modular units: macroblock coding unit, motion estimation unit, restructure from motion unit.
Make each block code be applicable to L1P, circulate take macro block group as unit at every turn, the size of macro block group is determined by L1D size.In macroblock coding module, when macro block group is sent in the process in sheet, reduce the size of data type as far as possible.Can just need not 16 bit data by 8 bit data, so not only save space, and can improve the service efficiency of L1D.Because the size that L1D is capable is fixed, if adopt 8 bit data in a line, can put one times than 16 bit data more, thus the generation of Cache deletion condition in minimizing program.
2), adopt ping-pong buffer structure, reduce the CPU stand-by period, as shown in Figure 2.
3), utilize EDMA to carry out data-moving, improve storage speed.
In USDC coded system structure, the EDMA function of DSP is in the situation that not having CPU to get involved, a kind of working method of reference to storage.It can directly pass through EDMA passage, in advance the data in peripheral hardware or chip external memory is directly moved to internal memory on sheet.Concerning DSP, the data of accessing are present in sheet, do not have the situation of data jamming to occur, and have reduced the CPU stand-by period.
4), use abundant video and image private library function.
Adopt the IMGLIB built-in function in CCS can efficiently realize motion estimation algorithm, the large SAD of amount of calculation and picture element interpolation have also been used to this instruction simultaneously, thereby improved the execution efficiency of software.
Although H.264 coding has carried out significantly compression to original image, limit by the network bandwidth, still can not carry out extensive remote video transmission, especially at the existing net environment of China.And H.264 USDC compression comparable little more than 4 times, and more clear.In other words, adopt exactly technical solution of the present invention in the situation that existing bandwidth condition is constant, to transmit the videos of at least 4 times more, even also can carry out transmission of video in the bad rural area of communication condition.
Should be understood that, application of the present invention is not limited to above-mentioned giving an example, and for those of ordinary skills, can be improved according to the above description or convert, and all these improvement and conversion all should belong to the protection range of claims of the present invention.
Claims (4)
1. a USDC data processing method, it comprises the following steps:
Comprise coloured image wavelet decomposition step, adopt supercomplex wavelet transformation, input picture is decomposed into the subgraph of different frequency, wherein low frequency subgraph similarly is the approximate of original image, the high frequency subgraph of three directions has reflected the detailed information of original image, retains the colouring information of coloured image;
Comprise vector coding step, adopt hypercomplex process of cluster analysis, retain human eye sensitive information content in image;
Comprise motion compensation step, adopt asymmetric cross multi-level hexagon lattice point to improve search, classify according to exercise intensity; Use the expansion of hexagon template simultaneously, the key character information of the temporal correlation between frame of video and image self is compressed, obtain the bit stream of compression;
Finally the bit stream of above-mentioned compression is exported from network.
2. USDC data processing method according to claim 1, is characterized in that, adopts macroblock coding unit to carry out coloured image wavelet decomposition step.
3. USDC data processing method according to claim 1, is characterized in that, adopts motion estimation unit to carry out vector coding step.
4. USDC data processing method according to claim 1, is characterized in that, adopts restructure from motion unit to carry out motion compensation step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410077400.2A CN103841421A (en) | 2014-02-26 | 2014-02-26 | USDC data processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410077400.2A CN103841421A (en) | 2014-02-26 | 2014-02-26 | USDC data processing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103841421A true CN103841421A (en) | 2014-06-04 |
Family
ID=50804500
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410077400.2A Pending CN103841421A (en) | 2014-02-26 | 2014-02-26 | USDC data processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103841421A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104735410A (en) * | 2015-02-11 | 2015-06-24 | 东电创新(北京)科技发展股份有限公司 | Narrow bandwidth lower than 4 K/S video transmission method and system |
| CN108629735A (en) * | 2017-03-24 | 2018-10-09 | 上海西门子医疗器械有限公司 | Detection image interpolation and the method and apparatus for estimating CT image resolution ratios |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080013846A1 (en) * | 2006-07-14 | 2008-01-17 | Sony Corporation | Wavelet transformation device and method, wavelet inverse transformation device and method, program, and recording medium |
| CN102547274A (en) * | 2010-11-23 | 2012-07-04 | 索尼公司 | Apparatus and method for reducing blocking artifacts |
-
2014
- 2014-02-26 CN CN201410077400.2A patent/CN103841421A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080013846A1 (en) * | 2006-07-14 | 2008-01-17 | Sony Corporation | Wavelet transformation device and method, wavelet inverse transformation device and method, program, and recording medium |
| CN102547274A (en) * | 2010-11-23 | 2012-07-04 | 索尼公司 | Apparatus and method for reducing blocking artifacts |
Non-Patent Citations (2)
| Title |
|---|
| 刘义鹏: "超复数描述下的超声图像压缩算法研究", 《中国优秀硕士学位论文全文数据库-信息科技辑》 * |
| 贺克军等: "H.264中运动估计算法研究与对比", 《广西工学院学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104735410A (en) * | 2015-02-11 | 2015-06-24 | 东电创新(北京)科技发展股份有限公司 | Narrow bandwidth lower than 4 K/S video transmission method and system |
| CN104735410B (en) * | 2015-02-11 | 2019-03-19 | 东电创新(北京)科技发展股份有限公司 | A kind of narrow bandwidth video transmission method and system lower than 4K/S |
| CN108629735A (en) * | 2017-03-24 | 2018-10-09 | 上海西门子医疗器械有限公司 | Detection image interpolation and the method and apparatus for estimating CT image resolution ratios |
| CN108629735B (en) * | 2017-03-24 | 2021-10-22 | 上海西门子医疗器械有限公司 | Method and device for detecting image interpolation and estimating CT image resolution |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Deepcoder: A deep neural network based video compression | |
| CN105659602B (en) | Coder side option for the intra block duplication prediction mode that video and image encode | |
| CN105684441B (en) | The Block- matching based on hash in video and image coding | |
| CN110024398A (en) | The estimation based on hash of the part of scene is remotely handled for screen | |
| TW202007166A (en) | Method of video decoding at a decoder and apparatus thereof | |
| CN108028919A (en) | Method and device for the context modeling of syntactic element in image and coding and decoding video | |
| US9497461B2 (en) | Method and apparatus for encoding frequency transformed block using frequency mask table, and method and apparatus for encoding/decoding video using same | |
| CN103782598A (en) | Fast encoding method for lossless coding | |
| CN105706450A (en) | Encoder decisions based on results of hash-based block matching | |
| CN107211139A (en) | The decoding escape pixel decoded for palette | |
| CN105659606A (en) | Features of base color index map mode for video and image coding and decoding | |
| CN105359531A (en) | Depth oriented inter-view motion vector prediction | |
| CN105765974A (en) | Features of intra block copy prediction mode for video and image coding and decoding | |
| CN110493596B (en) | Video coding system and method based on neural network | |
| CN104205835B (en) | The context of most end position decoding is exported for video coding | |
| CN103442228B (en) | Code-transferring method and transcoder thereof in from standard H.264/AVC to the fast frame of HEVC standard | |
| Jia et al. | Optimized inter-view prediction based light field image compression with adaptive reconstruction | |
| CN101984665A (en) | Video transmission quality evaluating method and system | |
| CN103347185A (en) | Unmanned aerial vehicle scouting image synthesis compressed encoding method based on selective block transformation | |
| KR102380579B1 (en) | Method and device for context-adaptive binary arithmetic coding of a sequence of binary symbols representing a syntax element related to video data | |
| CN107409212A (en) | The gradual renewal using conversion coefficient for coding and decoding | |
| Bernatin et al. | Video compression based on Hybrid transform and quantization with Huffman coding for video codec | |
| CN102137257B (en) | Embedded H.264 coding method based on TMS320DM642 chip | |
| CN115836525B (en) | Video encoding, decoding method and apparatus for prediction from multiple cross components | |
| CN110913232B (en) | Selection method and device of TU division mode and readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140604 |