CN105554501A - Image acquisition and compression method and device - Google Patents
Image acquisition and compression method and device Download PDFInfo
- Publication number
- CN105554501A CN105554501A CN201510914041.6A CN201510914041A CN105554501A CN 105554501 A CN105554501 A CN 105554501A CN 201510914041 A CN201510914041 A CN 201510914041A CN 105554501 A CN105554501 A CN 105554501A
- Authority
- CN
- China
- Prior art keywords
- image
- data
- compression
- matrix
- thumbnail
- 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.)
- Granted
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The present invention relates to an image acquisition and compression method and belongs to the digital image acquisition and processing field. The method includes the following steps that: the thumbnail data of an original image are acquired through downsampling; sensing matrix sampling is performed on the original image, so that the compression data of the original image can be obtained; the thumbnail image data and the compression data obtained through compressing a sensing matrix are outputted sequentially according to a preset compression ratio, so that codes streams which are to be transmitted or stored can be formed; and after the code streams are stored and transmitted, an image can be previewed through the thumbnail image data, and the original image is reconstructed based on the thumbnail image data and the compression data obtained through compressing the sensing matrix. With the method of the invention adopted, the complexity of a traditional image compression method can be effectively decreased; a defect of a method according to which an image can be seen through a reconstruction step based on conventional sensing data compression can be eliminated; the thumbnail image data are adopted as constraint conditions, the solution space of image reconstruction can be effectively reduced, and computing speed is improved; and a requirement for a network in compression data transmission can be decreased.
Description
Technical field
The present invention relates to a kind of data acquisition and compression method and device, particularly a kind of efficient image collection and compression method and device, belong to digital image acquisition and processing technology field.
Background technology
In order to solve the challenge of the required storage that faces and transmission when processing image mass data, traditional method normally adopts compress technique, namely expressing by realizing a kind of refining to raw digital signal, reducing the restriction of initial data to memory space and transmission bandwidth.Current compress technique is roughly divided into two classes: Lossless Compression and lossy compression method.Lossless Compression is as the term suggests be exactly that refining after utilizing Signal Compression is expressed and can be recovered raw digital signal without any distortion; Lossy compression method is then that the refining after utilizing Signal Compression is expressed and generally can be recovered primary signal, although signal and primary signal have certain error after recovering, error is in acceptable scope in specific applications.Clearly, Lossless Compression is very attractive, but the compression ratio that can provide due to it is limited, is not often suitable for the occasion needing large compression ratio.Data compression technique is almost ubiquitous, and such as, the picture of our shooting, the music of listening, the video of appreciation, even in star-loaded optical remote sensing field, nearly all O-E Payload is all furnished with special data compression unit.
Transform domain coding is a kind of comparatively popular data compression method, and it transforms to primary signal in some suitable transform domains usually, digs the openness expression of the number of it is believed that in this transform domain or compressible expression-form.Here " openness expression " refers to, suppose that original signal strength is N, this signal only has the coefficient of K non-zero in the transform domain as illustrated, and K N, utilize this K nonzero coefficient can express original input signal well." compressible expression-form " refers to that primary signal can be expressed by the coefficient of K non-zero well approx.Realized the compression of signal by the mode excavating signal openness expression, this compress mode adopt by many compression standards, such as JPEG, JPEG2000, H.264 with MP3 etc.Why signal can be because signal itself has very large redundancy by compression, no matter is that voice signal or picture signal are not always the case.We look back the conventional compress technique course of work: first realize the sampling of analog signal to digital signal, wherein containing a large amount of redundant datas, then excavate the openness of signal by transform domain again, realize compression finally by compression algorithm.This process contains huge waste in fact, first a large amount of redundant datas is gathered, then again these redundant datas are removed in compression process, so why not, abandon those redundant datas at the very start, the effective data of direct collection, so not only can save the cost of data acquisition, can also save space, this has just drawn the theory " compressed sensing " that this patent adopts.
The English of " compressed sensing " is expressed as Compressivesensing or Compressedsensing also or Compressivesampling, is abbreviated as CS.Simple " compression " this word, we are readily appreciated that, namely originally there being the data of redundancy to weed out, form the refined data more saving memory headroom; Simple " perception " this word is also readily appreciated that, i.e. signal sampling (analog signal becomes the process of digital signal)." compressed sensing " this blunt translation be not readily appreciated that at the beginning, but after we understand the theory of its behind, just slowly can understand its essence, namely compression and sampling are united two into one, the process that the process of namely sampling namely is compressed, the data after compressed perception sampling inherently compress after data.Compressive sensing theory points out that the signal of limited dimension that is sparse or that have sparse expression can utilize the measured value that is linear, non-self-adapting far fewer than nyquist sampling quantity to rebuild out undistortedly.For a signal x ∈ R
n, in x, only comprise K nonzero value.Suppose that we obtain M linear measurement by a perception matrix, namely we can describe this sampling process y=x by Mathematical Modeling below, be wherein a size are the matrix of M*N, the measured value of gained of namely sampling.The projection operation of a matrix notation dimensionality reduction, R
nbe mapped to R
min, in general K<M N, namely matrix column number is far more than line number, and this mathematical notation is namely to standard compression perception frame description.This theory, once proposition, causes extensive concern in numerous areas such as information theory, signal/image procossing, imaging of medical, radio astronomy, pattern recognition, optics/radar imagery, chnnel coding etc.
Although we are known can complete collection to echo signal and compression very efficiently by compression sensing method, but routine has to pass through reconstruction procedures based on the data that compressed sensing obtains could recover primary signal, and utilize the optimized algorithm of L1 norm minimum to need a large amount of calculating, significant discomfort closes handheld device, so mean based on the camera of compression sensing method and our are conventional uses digital camera to be accustomed to disagreing, because cannot check shooting effect by playback in real time.
Summary of the invention
The object of the invention is to solve the complicated poor efficiency of traditional images collection and compression method, the compression sensing method problem opposing with the traditional use habit of user, a kind of IMAQ and compression method are efficiently provided, direct collection thumbnail data and packed data, both met the demand that user reviews imaging effect immediately, and original image can have been rebuild when user needs to check original image details again.
Object of the present invention is achieved through the following technical solutions:
A kind of IMAQ and compression method, comprise following content:
The thumbnail data that down-sampling obtains original image is carried out to the original image that pattern matrix detector obtains;
The packed data that perception matrix sampling obtains original image is carried out to the original image that pattern matrix detector obtains;
In conjunction with thumbnail data and packed data as the code stream stored or transmit;
Utilize the thumbnail preview image in code stream;
In conjunction with thumbnail data and packed data reconstituting initial image.
As preferably, the byte number of described thumbnail data and described packed data can flexible configuration, under the prerequisite of fixing compression ratio, the byte number of thumbnail data is determined according to down-sampling multiple, and in fixing code stream required by compression ratio, remaining byte number is filled by packed data.
As preferably, described perception matrix is meet or approximately meet in compressive sensing theory the matrix retraining equidistant characteristics (restrictedisometryproperty, RIP).
As preferably, described matrix is noise waves transformation matrix (Noiselettransformmatrix), Gaussian matrix (Gaussianmatrix), Bei Nuli matrix (Bernoullimatrix), part fourier transform matrix (PartialFouriertransformmatrix) or part hadamard matrix (PartialHadamardmatrix).
As preferably, in order to improve compression ratio further, before carrying out described code stream storage or transmission, first lossless compression-encoding is adopted to encode to code stream.
As preferably, described lossless compression-encoding is Huffman encoding (Huffmancoding), the coding (Arithmeticcoding) that counts, run length encoding (Run-LengthEncoding) or self-adapting dictionary coding (adaptivedictionaryencoding).
As preferably, described in conjunction with thumbnail data and the packed data that obtains through perception matrix as the code stream stored or transmit, its order is for first to store or transmission thumbnail data stores or transmits the packed data obtained through perception matrix.
As preferably, described in conjunction with thumbnail data and packed data reconstituting initial image, adopt dig according to original image in the transform domain as illustrated L1 norm minimum optimization method realize, and should equal or be approximately equal to the thumbnail data of receiving terminal or storage end as constraints by the down-sampled data of reconstructed image, thus reach the object reducing reconstructed image solution space.
A kind of IMAQ and compression set, comprise image detection module, down sample module, compressed sensing module, packed data output module, image reconstruction module; Wherein image detection module is connected with compressed sensing module with down sample module respectively, and down sample module is connected with packed data output module respectively with compressed sensing module;
Described image detection module is for obtaining original image;
Described down sample module is used for carrying out down-sampling to the raw image array that image detection module exports and obtains thumbnail data;
Described compressed sensing module is used for carrying out obtaining packed data based on the perception matrix sampling of compressive sensing theory to the original image that image detection module exports;
The packed data that described packed data output module exports for the thumbnail data that exports in conjunction with down sample module and compressed sensing module is as the code stream for storing or transmit and export;
Described image reconstruction module is used for rebuilding original image according to thumbnail data and packed data.
As preferably, described compressed sensing module adopts noise waves conversion to realize, and described image reconstruction module adopts the optimization method dug according to original image L1 norm minimum in the transform domain as illustrated to realize.
Beneficial effect
Efficient image collection provided by the invention and compression method, compared to existing technologies, on the one hand, be used as the down-sampling thumbnail of original image as packed data; On the other hand, the compression to original image is realized by compressed sensing matrix.The advantage of this mode is: first, solve based under compressive sensing theory imaging system, immediately the problem of imaging effect cannot be reviewed, because the method merely based on compressive sensing theory imaging needs one usually based on the process of reconstruction of L1 norm minimum, and the present invention realizes immediately reviewing imaging effect by checking the mode of thumbnail; The second, the down-sampling thumbnail of original image is used as constraints, the reconstructed image quality based on L1 norm minimum is significantly improved; 3rd, significantly reduce the complexity of IMAQ and compression.
The present invention is the packed data being then only compressed perception acquisition with prioritised transmission thumbnail at coding or transmission sequence, according to the benefit of this Sequential output be, on the one hand, first receiving terminal sees thumbnail, can select whether to be necessary super-resolution reconstruction original image; On the other hand, according to this transmission means, allow receiving terminal after receiving thumbnail under any compression ratio interrupting receive, do not affect the recovery of complete image, just difference to some extent in reconstructed image quality, namely receives more that multiple pressure contracting data reconstruction picture quality is better.
Accompanying drawing explanation
Fig. 1 is a kind of IMAQ of the embodiment of the present invention and compression method schematic flow sheet;
Fig. 2 is a kind of IMAQ of the embodiment of the present invention and compression set structural representation.
Reference numeral: 101-pattern matrix detector; 102-down sample module; 103-compressed sensing module; 104-packed data output module; 105-image reconstruction module.
Embodiment
Below in conjunction with accompanying drawing and embodiment, the present invention is described in detail.
Embodiment 1
Below for the quantization resolution of each pixel be 8 bits, fixing compression ratio is 8 to be described in detail to a kind of IMAQ of the present invention's proposition and compression method, as shown in Figure 1, comprises following content:
1, the original image obtained pattern matrix detector carries out the thumbnail data that down-sampling obtains original image.
Cmos image detector array and 256 times of down-samplings explanation down-sampling processes so that array size is 2048 × 2048: the fritter being divided into several virtual by 16 × 16 this array, according to from left to right, order from the top down only reads out a pixel in the most upper left corner in each fritter, so just achieve 256 times of down-samplings to original image, obtain thumbnail data.According to this example, those skilled in the art can adopt pair array in the same way and carry out piecemeal and the down-sampling choosing arbitrarily the required multiple of pixel realization of each piecemeal; Certain pattern matrix detector directly can also select existing General Cell detector.
For this process, formally y can be expressed as
1=Dx, wherein, vector x represents original image, and matrix D represents this above-mentioned down-sampling process, vector y
1represent the thumbnail obtained after down-sampling.Owing to being 256 times of down-samplings, therefore output vector y
1byte number be 128 × 128=16384.Basically this step is a process read, and does not change the quantized values of pixel on detector array in essence.
2, the original image obtained pattern matrix detector carries out the packed data that perception matrix sampling obtains original image.
Because fixing compression ratio is 8, the final data word joint number exported should be 2048*2048/8=524288; Owing to having obtained the thumbnail data of 16384 bytes through down-sampling, so the data of 524288-16384=507904 byte can only be exported after perception matrix sampling.Can be understood by this process, the byte number of thumbnail data and packed data can flexible configuration, under the prerequisite of fixing compression ratio, the byte number of thumbnail data is determined according to down-sampling multiple, and in fixing code stream required by compression ratio, remaining byte number is filled by packed data.
Matrix in describing before just achieving in conjunction with the known perception matrix sampling of background technology, the projection operation of a perception matrix notation dimensionality reduction, R
nbe mapped to R
min, in general K<M N, namely matrix column number is far more than line number, conventional perception matrix comprises: Gaussian matrix, Bei Nuli matrix and part fourier transform matrix, but considering the embody rule that practical feasibility may limit they, Gauss's sampling matrix of such as Gaussian distributed is the feasibility almost do not realized at hardware, on the one hand, we cannot store the sampling matrix of this Gaussian Profile that obeys without questioning in memory; On the other hand, conventional memory cannot the huge sampling matrix of storage size (when we need according to a number of samples M=10000 rebuild one there is the image N=1000000 of mega pixel time, we just need the internal memory of 10G byte, this is very unpractical in the application of reality, and also do not consider amount of calculation required for process of reconstruction here, so adopt completely random measurement to be very unpractiaca in the application of reality).Therefore, the present embodiment realizes perception matrix sampling so that the noise waves of software and hardware realization is transformed to example.But skilled in the art will recognize that, according to the principle of compressed sensing technology, everyly meet or approximate meet in compressive sensing theory the matrix retraining equidistant characteristics and can be applied to this, as noise waves transformation matrix (Noiselettransformmatrix), Gaussian matrix (Gaussianmatrix), Bei Nuli matrix (Bernoullimatrix), part fourier transform matrix (PartialFouriertransformmatrix) and part Hadamard (PartialHadamardmatrix).
Use the process of perception matrix sampling as follows:
Can utilize to the pixel of whole pattern matrix do noise waves conversion more random output noise wave system number realize, this computational process, can with reference to real number noise waves transform method (R.Coifman, F.GeshwindandY.Meyer, " Noiselets ", AppliedandComputationalHarmonicAnalysis10, 2001), concrete hardware implementing can adopt field programmable gate array FPGA(Field-ProgrammableGateArray) or other logic control device carry out this conversion process of hardware implementing, but in order to save power consumption, the object of volume and weight, final this hardware (can realize the module of perception matrix sampling function) is integrated into the form of flow on pattern matrix detector.This conversion is a general conversion, and no matter which type of scene image incident is through this conversion, and the data after reading will become the same signal of similar noise, and the noise like coefficient after conversion is similarly the size of original image, is 2048 × 2048.507904 bytes are chosen as y randomly from the coefficient after conversion
2.Although wherein the reading of this random site is before flow or stochastic generation before hardware implementing, but once be constant after hardware solidification, as the random number functions that we often use, as long as initial condition is the same, the random number sequence that each run obtains must be the same.Such as: the value at every turn exported after compressed sensing module is all position (1,3), (71,29), (41,79), (32,68) ...Noise waves coefficient on these positions is read the conversion the acquisition completing information that can complete matrix successively according to a fixing sequence, and this process formally can be expressed as y
2=x, wherein, vector x represents original image, matrix notation said sensed matrix sampling process, vector y
1represent the packed data obtained after perception matrix sampling.
3, in conjunction with thumbnail data and packed data as the code stream stored or transmit.
Undertaken above-mentioned thumbnail data and packed data encoding the code stream exporting and namely formed and be used for transmitting or storing, and according to the use habit of user, the present embodiment adopts thumbnail data in the posterior sequential storage of front packed data or transmission, is formally expressed as [y
1, y
2], can be understood as approximate Embedded coding form.According to the benefit of this Sequential output be: on the one hand, important measured value y
1preferential transmission, first receiving terminal sees thumbnail, can select whether to be necessary super-resolution reconstruction original image, and when unnecessary, transmitting terminal is without the need to sending packed data y
2, can bandwidth be saved; On the other hand, according to this transmission means, allow receiving terminal after receiving thumbnail, under any compression ratio, interrupt the reception of packed data, do not affect the recovery of complete image, be according to the size of packed data difference to some extent in reconstructed image quality, namely the larger reconstructed image quality of compression ratio is poorer, and the result of the less reconstruction of compression ratio is more close to original input picture x.
In addition, if in order to improve compression ratio further, additionally can also embed traditional lossless compression-encoding method, such as Huffman encoding (Huffmancoding), the coding (Arithmeticcoding) that counts, run length encoding (Run-LengthEncoding), self-adapting dictionary coding (adaptivedictionaryencoding) etc.
4, the thumbnail preview image in code stream is utilized; In conjunction with thumbnail data and packed data reconstituting initial image.
The data be through after perception matrix compression because packed data is not through image data, so the process must rebuild through could recover original image.Because the sampled value number obtained through perception matrix sampling is obviously less than the number of pixels of original image, namely the number of unknown number is far fewer than the number of equation, so this kind of reconstruction is an ill-conditioning problem, often there is numerous solution and meets equation.Solve this kind of ill-conditioning problem, the most classical method is no more than maximum a posteriori probability (MaximumAPosteriori, MAP), but often need the prior probability model of an echo signal as a constraints, thus from numerous solution, just likely find that solution closest to target image.If add that other qualifications can also reduce solution space further, so using the result of thumbnail as original image down-sampling in the present invention, this constraints will play very important effect in the process solved by optimized algorithm.
Therefore by can by the direct preview image of thumbnail data after code stream decoding; By digging the optimization method reconstituting initial image according to original image L1 norm minimum in the transform domain as illustrated, namely following cost function (CostFunction) is solved:
Min{||W(x)||
1}s.t.y
1=Dx；
y
2=??x；
Wherein, s.t. represents constraints, and constraints is 2 herein, is respectively y
1=Dx and y
2=x, implication is based on packed data y
2the down-sampling of the image x of reconstruct should equal receiving terminal or store the thumbnail data of end; || ||
1represent L
1norm; Function W () presentation video x can embody openness in W transform domain, such as W can be wavelet transformed domain, this is because it is openness to it has been generally acknowledged that natural image can embody in wavelet field, but when W is selected, be not limited thereto, and depend on the degree of understanding to image sparse, i.e. priori.Solving of above-mentioned cost function can adopt regulatory thresholds well known to those skilled in the art to shrink iterative algorithm (IterativeShrinkage-ThresholdingAlgorithmfor, ISTA) realization.
Embodiment 2
According to embodiment 1, IMAQ and compression method achieve a set of IMAQ and compression set, structure as shown in Figure 2, comprising: image detection module 101, down sample module 102, compressed sensing module 103, packed data output module 104, image reconstruction module 105; Wherein image detection module is connected with compressed sensing module with down sample module respectively, and down sample module is connected with packed data output module respectively with compressed sensing module;
Described image detection module, for obtaining original image, adopts existing pattern matrix detector to realize in the present embodiment;
Described down sample module is used for carrying out down-sampling to the original image that image detection module exports and obtains thumbnail data;
Described compressed sensing module is used for carrying out obtaining packed data based on the perception matrix sampling of compressive sensing theory to the original image that image detection module exports, and adopts IMAQ described in embodiment 1 and the noise waves described in compression method to convert and realize in the present embodiment;
The packed data that described packed data output module exports for the thumbnail data that exports in conjunction with down sample module and compressed sensing module is as the code stream for storing or transmit and export, and adopts the order of packed data after first thumbnail data in IMAQ described in embodiment 1 and compression method to generate code stream in the present embodiment;
Described image reconstruction module is used for rebuilding original image according to thumbnail data and packed data, adopts in IMAQ described in embodiment 1 and compression method the optimization method dug according to original image L1 norm minimum in the transform domain as illustrated to realize in the present embodiment.
In sum, the method that the present invention combines with compressed sensing model by adopting down-sampling, can meet handheld device user checks shooting effect requirement by thumbnail.Moreover, play the part of the role of constraints when thumbnail also utilizes L1 norm minimum to rebuild original image in this patent, the reconstructing method adding constraints so this is much better than the picture quality based on conventional compact sensing reconstructing under same compression ratio.
And, adopt the inventive method also to eliminate in routine variations coding and find important conversion coefficient and complex process to important transform coefficients encoding, thus have important practical significance: such as, for spaceborne optical imaging apparatus, omit whole compression unit to mean and save a large amount of power consumption, volume, this is concerning significant space remote sensing.This method also agrees with civil camera low-power consumption, light active demand very much simultaneously.The difficult problem all the time perplexing handheld camera or mobile phone camera manufacturer is exactly the power problems of digital camera.Such as, the peak value total power consumption of JPEG2000 compressed encoding chip ADV202 that U.S. ADI chip companies is produced can reach 0.9 watt nearly, and this is concerning the handheld device more and more popularized the present age being really a very large burden.Therefore can reduce the power consumption of handheld device by this patent method and extend its battery working time, the volume and weight of handheld device can certainly be reduced.
Above-described specific descriptions; the object of inventing, technical scheme and beneficial effect are further described; be understood that; the foregoing is only specific embodiments of the invention; the protection range be not intended to limit the present invention; within the spirit and principles in the present invention all, any amendment made, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. IMAQ and a compression method, is characterized in that, comprises following content:
The thumbnail data that down-sampling obtains original image is carried out to the original image that pattern matrix detector obtains;
The packed data that perception matrix sampling obtains original image is carried out to the original image that pattern matrix detector obtains;
In conjunction with thumbnail data and packed data as the code stream stored or transmit;
Utilize the thumbnail preview image in code stream;
In conjunction with thumbnail data and packed data reconstituting initial image.
2. a kind of IMAQ according to claim 1 and compression method, it is characterized in that, the byte number of described thumbnail data and described packed data can flexible configuration, under the prerequisite of fixing compression ratio, the byte number of thumbnail data is determined according to down-sampling multiple, and in fixing code stream required by compression ratio, remaining byte number is filled by packed data.
3. a kind of IMAQ according to claim 1 and compression method, is characterized in that, described perception matrix is meet or be similar to meet in compressive sensing theory the matrix retraining equidistant characteristics.
4. a kind of IMAQ according to claim 3 and compression method, is characterized in that, described matrix is noise waves transformation matrix, Gaussian matrix, Bei Nuli matrix, part hadamard matrix or part fourier transform matrix.
5. a kind of IMAQ according to claim 1 and compression method, is characterized in that, in order to improve compression ratio further, before storing described code stream or transmitting, adopts lossless compression-encoding to encode to code stream.
6. a kind of IMAQ according to claim 5 and compression method, is characterized in that, described lossless compression-encoding is Huffman encoding, count coding, run length encoding or self-adapting dictionary coding.
7. a kind of IMAQ according to claim 1 and compression method, it is characterized in that, described in conjunction with thumbnail data and the packed data that obtains through perception matrix as the code stream stored or transmit, its order be first to store or transmission thumbnail data recompresses data.
8. a kind of IMAQ according to claim 1 and compression method, it is characterized in that, described in conjunction with thumbnail data and packed data reconstituting initial image, adopt dig according to original image in the transform domain as illustrated L1 norm minimum optimization method realize, and should equal or be approximately equal to the thumbnail data of receiving terminal or storage end as constraints by the down-sampling of reconstructed image, thus reach the object reducing reconstructed image solution space.
9. IMAQ and a compression set, is characterized in that, comprises image detection module, down sample module, compressed sensing module, packed data output module, image reconstruction module; Wherein image detection module is connected with compressed sensing module with down sample module respectively, and down sample module is connected with packed data output module respectively with compressed sensing module;
Described image detection module is for obtaining original image;
Described down sample module is used for carrying out down-sampling to the raw image array that image detection module exports and obtains thumbnail data;
Described compressed sensing module is used for carrying out obtaining packed data based on the perception matrix sampling of compressive sensing theory to the original image that image detection module exports;
The packed data that described packed data output module exports for the thumbnail data that exports in conjunction with down sample module and compressed sensing module is as the code stream for storing or transmit and export;
Described image reconstruction module is used for rebuilding original image according to thumbnail data and packed data.
10. a kind of IMAQ according to claim 9 and compression set, it is characterized in that, described compressed sensing module adopts noise waves conversion to realize, and described image reconstruction module adopts the optimization method dug according to original image L1 norm minimum in the transform domain as illustrated to realize.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510914041.6A CN105554501B (en) | 2015-05-03 | 2015-12-12 | Image acquisition and compression method and device |
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510216258.XA CN104811621A (en) | 2015-05-03 | 2015-05-03 | High-efficiency image collection and compression method |
| CN201510216258X | 2015-05-03 | ||
| CN201510914041.6A CN105554501B (en) | 2015-05-03 | 2015-12-12 | Image acquisition and compression method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105554501A true CN105554501A (en) | 2016-05-04 |
| CN105554501B CN105554501B (en) | 2020-04-24 |
Family
ID=53696086
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510216258.XA Pending CN104811621A (en) | 2015-05-03 | 2015-05-03 | High-efficiency image collection and compression method |
| CN201510914041.6A Active CN105554501B (en) | 2015-05-03 | 2015-12-12 | Image acquisition and compression method and device |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510216258.XA Pending CN104811621A (en) | 2015-05-03 | 2015-05-03 | High-efficiency image collection and compression method |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN104811621A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106792272A (en) * | 2016-11-28 | 2017-05-31 | 维沃移动通信有限公司 | The generation method and mobile terminal of a kind of video thumbnails |
| CN107277521A (en) * | 2016-09-06 | 2017-10-20 | 芯原微电子(北京)有限公司 | A kind of TU trees choice algorithm based on HEVC Video codings |
| CN107515227A (en) * | 2017-07-27 | 2017-12-26 | 武汉精测电子技术股份有限公司 | Based on the lossless method for compressing image of defect and defects of display panel detecting system |
| CN109495113A (en) * | 2018-10-19 | 2019-03-19 | 华南理工大学 | A kind of compression method and device of EEG signals |
| CN110148087A (en) * | 2019-05-22 | 2019-08-20 | 杭州电子科技大学 | Compression of images and method for reconstructing based on rarefaction representation |
| CN111130553A (en) * | 2018-10-31 | 2020-05-08 | 罗伯特·博世有限公司 | Method and apparatus for processing compressed data |
| CN111881942A (en) * | 2020-07-02 | 2020-11-03 | 中国空间技术研究院 | Target classification method and system based on compressed learning |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105227839B (en) * | 2015-10-08 | 2018-05-08 | 杭州晟元数据安全技术股份有限公司 | A kind of efficient image acquisition hardware controller |
| CN105451024B (en) * | 2015-12-31 | 2021-01-26 | 北京大学 | Digital hologram coding transmission method adopting compressed sensing |
| CN110554367B (en) * | 2018-06-04 | 2023-06-06 | 重庆测威科技有限公司 | Target scattering characteristic measurement interference removal method based on compressed sensing |
| CN109816542B (en) * | 2019-01-28 | 2023-04-07 | 中国平安财产保险股份有限公司四川分公司 | Method and system for reducing yield and settling claim of crops |
| CN113689330A (en) * | 2021-07-13 | 2021-11-23 | 中国空间技术研究院 | Image reconstruction method and device and image compression method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103186891A (en) * | 2013-03-11 | 2013-07-03 | 西安电子科技大学 | Hexagon image reconstruction method based on compressed sensing |
| CN103747259A (en) * | 2013-12-27 | 2014-04-23 | 华为技术有限公司 | Video processing method and device |
-
2015
- 2015-05-03 CN CN201510216258.XA patent/CN104811621A/en active Pending
- 2015-12-12 CN CN201510914041.6A patent/CN105554501B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103186891A (en) * | 2013-03-11 | 2013-07-03 | 西安电子科技大学 | Hexagon image reconstruction method based on compressed sensing |
| CN103747259A (en) * | 2013-12-27 | 2014-04-23 | 华为技术有限公司 | Video processing method and device |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107277521A (en) * | 2016-09-06 | 2017-10-20 | 芯原微电子(北京)有限公司 | A kind of TU trees choice algorithm based on HEVC Video codings |
| CN106792272A (en) * | 2016-11-28 | 2017-05-31 | 维沃移动通信有限公司 | The generation method and mobile terminal of a kind of video thumbnails |
| CN107515227A (en) * | 2017-07-27 | 2017-12-26 | 武汉精测电子技术股份有限公司 | Based on the lossless method for compressing image of defect and defects of display panel detecting system |
| CN109495113A (en) * | 2018-10-19 | 2019-03-19 | 华南理工大学 | A kind of compression method and device of EEG signals |
| CN111130553A (en) * | 2018-10-31 | 2020-05-08 | 罗伯特·博世有限公司 | Method and apparatus for processing compressed data |
| CN111130553B (en) * | 2018-10-31 | 2024-05-24 | 罗伯特·博世有限公司 | Method and apparatus for processing compressed data |
| CN110148087A (en) * | 2019-05-22 | 2019-08-20 | 杭州电子科技大学 | Compression of images and method for reconstructing based on rarefaction representation |
| CN110148087B (en) * | 2019-05-22 | 2023-08-04 | 湖南华凯数字科技有限公司 | Image compression and reconstruction method based on sparse representation |
| CN111881942A (en) * | 2020-07-02 | 2020-11-03 | 中国空间技术研究院 | Target classification method and system based on compressed learning |
| CN111881942B (en) * | 2020-07-02 | 2023-09-26 | 中国空间技术研究院 | Target classification method and system based on compression learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104811621A (en) | 2015-07-29 |
| CN105554501B (en) | 2020-04-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105554501A (en) | Image acquisition and compression method and device | |
| Yang et al. | An overview of lossless digital image compression techniques | |
| Jasmi et al. | Comparison of image compression techniques using huffman coding, DWT and fractal algorithm | |
| CN109600618A (en) | Video-frequency compression method, decompression method, device, terminal and medium | |
| JP2015504545A (en) | Predictive position coding | |
| Chawla et al. | Image compression techniques: a review | |
| CN111050170A (en) | Image compression system construction method, compression system and method based on GAN | |
| CN112887713B (en) | Picture compression and decompression method and device | |
| CN104869426A (en) | JPEG coding method lowering image diamond effect under low compression code rate | |
| CN105828087A (en) | Image compression system based on optical wavelet and compressed sensing | |
| Kim et al. | Hierarchy embedded differential image for progressive transmission using lossless compression | |
| Tilton et al. | Earth science data compression issues and activities | |
| Bhade et al. | Comparative study of DWT, DCT, BTC and SVD techniques for image compression | |
| CN115294222A (en) | Image encoding method, image processing method, terminal, and medium | |
| CN104079930A (en) | Achieving method of remote-sensing image compression system | |
| CN113096019B (en) | Image reconstruction method, image reconstruction device, image processing equipment and storage medium | |
| CN113949867B (en) | Image processing method and device | |
| Garg | An improved algorithm of fractal image compression | |
| Singh et al. | A framework to design novel SVD based color image compression | |
| Phadatare et al. | Implementation of DCT and Fractal Compression Technique | |
| Ghamisi et al. | A new method for compression of remote sensing images based on an enhanced differential pulse code modulation transformation | |
| CN115052148B (en) | Image compression algorithm based on model segmentation compression self-encoder | |
| Bhuvaneswari et al. | Fast fractal image compression algorithm based image interfacing system | |
| Das et al. | Image compression using discrete cosine transform & discrete wavelet transform | |
| Singh et al. | COMPARATIVE STUDIES OF VARIOUS TECHNIQUES FOR IMAGE COMPRESSION ALGORITHM |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |