CN103886106B - Remote sensing image safe-retrieval method based on spectral feature protection - Google Patents
Remote sensing image safe-retrieval method based on spectral feature protection Download PDFInfo
- Publication number
- CN103886106B CN103886106B CN201410149032.8A CN201410149032A CN103886106B CN 103886106 B CN103886106 B CN 103886106B CN 201410149032 A CN201410149032 A CN 201410149032A CN 103886106 B CN103886106 B CN 103886106B
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- image
- matrix
- sensing images
- retrieval
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/602—Providing cryptographic facilities or services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/58—Extraction of image or video features relating to hyperspectral data
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Library & Information Science (AREA)
- Computer Security & Cryptography (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioethics (AREA)
- General Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of image processing and particularly relates to a remote sensing image safe-retrieval method based on spectral feature protection. To guarantee the safety of the retrieval process of remote sensing images, the remote sensing image retrieval method of an encrypted domain is provided on the basis of the prominent characteristics of specific spectral features of the remote sensing images. After the spectral features of the images are extracted, an encryption protection method for the spectral features is added, under the condition of no decryption, image similarity measure is directly performed on encrypted feature vectors, and thus the safety of image information is guaranteed while the retrieval function is completed. A basis is laid for remote sensing image target recognition and image classification and retrieval.
Description
Technical field
The invention belongs to technical field of image processing is and in particular to a kind of remote sensing images safety based on spectral signature protection
Search method, can be used for remote Sensing Image Retrieval research.
Background technology
Remote sensing be a kind of using electromagnetic wave from remote technology and the method that earth's surface is carried out with noncontact detection.Its basis reason
By being, because the species of earth's surface and residing environment are different, thus have different to the electromagnetic wave signal of different wave length
Reflection or radiation effect.Remote sensing technology exactly utilizes this speciality, by the electromagnetic wave information of the different earth's surface of research, thus analyzing
Its physical characteristic and geological information.In recent years, the correlation technique such as sensor technology, remote sensing platform technology, data communication technology
Obtain development at full speed, modern Remote Sensing Technical has come into one and can dynamic, quick, accurate, timely, many means provide
The new stage of multiple earth observation data, become people and obtain environmental information, an important means of epistemic context.
Remote sensing images are from the workbench away from ground, by sensor, the electromagnetic wave information of earth surface are carried out
Detect, then again through information transfer, process and analysis, finally give remote sensing images.It has very high spatial resolution and
Spectral resolution, can obtain ground object target image in hundreds of or even thousands of continuous spectrum wave bands.With people to remote sensing
The dependent of image is gradually increased, and the process of such image has also become focus of concern.At present, remote sensing images become
The capsule information in the fields such as precision agriculture, atmospheric seeing, environmental monitoring, disaster study, resource investigation.Remote Sensing Image Retrieval
It is not that image retrieval technologies directly, are simply applied in remote sensing fields, but is related to some new key technologies, need by one
Solving, many correlative studys at present still need to be further improved and ripe for new method and thinking a bit.
Remote Sensing Image Retrieval technology is also applied to increasing field, especially military field therewith, and remote sensing images exist
Understanding many aspects such as battlefield surroundings and Hitting Effect Evaluation are applied.On the other hand, had by cybertimes
Open with sharing so that people, while enjoying the convenience that present information brought, the huge interests such as shared, also face
Various information security threats, such as remote sensing images the safety problem such as illegally usurp, distort, steal and check, cause
Greatly insecurity.Common remote sensing images are stolen or check may impact less, and for containing important information
Remote sensing images, how to ensure its retrieval security become domestic and international researcher focus of interest, have very high
Researching value.
Remote sensing images, especially military or business remote sensing images are related to national security or military secrecy, therefore to distant
The retrieval security requirement of sense image is very high.Encryption technology is a kind of effective means ensureing safety of image retrieval, that is, to remote sensing
Image information is encrypted and becomes ciphertext, legitimate receiver according to authorize the ciphertext receiving can be decrypted recover secret
Confidential information.And these information are not to knowing that the network interception person of key is unrecognizable, and then can effectively protect retrieval
During view data safety.A kind of effective safety image search index scheme, needs to take into account validity and extensibility,
Be able to maintain that image between similitude, and guarantee safety using suitable key.Therefore, in the situation that key is unknown
Under, the information of search database or deduction data-base content is highly difficult.So can effectively ensure that remote sensing images are examined
Security during rope.
Content of the invention
The present invention is different from existing remote sensing image retrieval method, the spy projecting for the distinctive spectral signature of remote sensing images
Point is it is proposed that a kind of remote sensing image retrieval method of encrypted domain.This technology is encrypted protection to the feature of remote sensing images, not
In the case of deciphering, directly the characteristic vector of encryption is carried out with the measuring similarity of image, while realizing retrieval performance, protects
The security of barrier information.On the basis of comprehensive analysis are carried out to remote sensing images, with the randomized method of bit-planes to figure
As feature is encrypted.During retrieval, feature need not be decrypted, directly similarity be carried out to the feature after encryption
Joining, using the image of feature similarity as retrieval result, realizing the safe retrieval of remote sensing images it is characterised in that including following walking
Suddenly:
Step 1:Preprocessing of remote sensing images
First deblocking is carried out to remote sensing images, the target image of input is divided into from left to right, from top to bottom rule,
Not overlapping and equal-sized subimage block (256*256,128*128,64*64,32*32,16*16).
Step 2:Remote-sensing image spectrum feature extraction
Using the spectral signature separating the method extraction remote sensing images that (MNF) converts based on minimal noise, method is as follows:
Step 2.1:Using high-pass filter, noise extraction is carried out to image, obtain noise covariance matrix CN, and by its
Diagonally turn to matrix DN, method is as follows:
DN=UTCNU (1)
Wherein, U is the orthogonal matrix being made up of characteristic vector;
Step 2.2:Original image is projected to new space, method is as follows:
I=PTCNP (2)
Step 2.3:Dialogue noise data carries out standard principal component transform, and method is as follows:
CT=PTCDP (3)
Step 2.4:Diagonally turn to matrix DT, method is as follows:
DT=VTCTV (4)
Step 2.5:Calculate minimal noise separation transformation matrix, method is as follows:
TMNF:TMNF=PV (5)
Wherein I is unit battle array, and P is transformation matrix, CDCovariance matrix for image X;CTIt is the matrix after P conversion,
DTFor CTCharacteristic value according to descending diagonal matrix, V is the orthogonal matrix being made up of characteristic vector.
Step 3:The encipherment protection of remote-sensing image spectrum feature
Process remote sensing images using the randomized method of bit-planes, remote-sensing image spectrum feature is encrypted, retain
The distance between original characteristic vector.
Step 4:The similitude coupling of spectral signature
Set up index, carry out remote sensing images coupling using the discrete matching method of spectral information, realize the safety inspection of remote sensing images
Rope.
The present invention compared with prior art, has following obvious advantage and beneficial effect:
The present invention is different from existing remote sensing image retrieval method, the spy projecting for the distinctive spectral signature of remote sensing images
Point is it is proposed that a kind of remote sensing image retrieval method of encrypted domain.This technology is encrypted protection to the feature of remote sensing images, not
In the case of deciphering, directly the characteristic vector of encryption is carried out with the measuring similarity of image, while realizing retrieval performance, protects
The security of barrier information.
Brief description
Fig. 1 piecemeal schematic diagram;
Fig. 2 minimal noise separation conversion process;
Fig. 3 remote sensing images piecemeal schematic diagram;
The encryption schematic diagram of j-th bit plane of Fig. 4;
The flow chart of Fig. 5 characteristics of image encryption;
Fig. 6 remote sensing images safe retrieval flow chart;
Fig. 7 process of feedback figure.
Specific embodiment
Below in conjunction with accompanying drawing, the present invention will be further described, and idiographic flow of the present invention is as follows:
Step 1:Preprocessing of remote sensing images
The present invention is primarily based on not overlapping region and carries out deblocking to high-spectrum remote sensing, using partitioned mode to distant
Sense image enters piecemeal and processes.One width remote sensing images are divided into regular, not overlapping and equal-sized son from left to right, from top to bottom
Image block.The indexed mode of piecemeal:If representing row, column number in raw video for certain subimage block with (row, col),
TileWidth and TileHeight represents the size of subimage block, then coordinate (the x in this subimage block upper left corner and the lower right cornermin,
ymin) and (xmax,ymax), method is as follows:
As shown in Figure 3, the size downloaded from AVIRIS high-spectrum remote sensing data storehouse by a width is 512*512 pixel
True color image illustrates piecemeal effect, and size is 8*8.This deblocking organizational form is very simple, directly perceived, easy to understand,
And be easy to the detection of Small object and piecemeal remote sensing image data is managed using data base management system.
Step 2:The extraction of remote-sensing image spectrum feature
The present invention separates MNF to spectroscopic data using minimal noise first and carries out leading dimensionality reduction to it.Will be important in image
Feature focuses in minority principal component;And then using a kind of method of feature based value, the principal component after conversion is intercepted,
Obtain the principal component that wherein information content maximum, energy are concentrated.
Step 2.1:Image is filtered process using high-pass filter, obtains noise covariance matrix CN, and by its
Diagonally turn to matrix DN:
DN=UTCNU (2)
Wherein, DNFor CNCharacteristic value according to descending diagonal matrix;U is the orthogonal matrix being made up of characteristic vector.Enter
One step conversion:
I=PTCNP (3)
Wherein, I is unit battle array;P is transformation matrix,
When P is applied to view data X, general Y=PX conversion, original image can be projected to new space, at this moment
Noise in conversion data has unit variance, and uncorrelated between wave band.
Step 2.2:Dialogue noise data carries out standard principal component transform:
CT=PTCDP (4)
Wherein, CDCovariance matrix for image X;CTIt is the matrix after P conversion, can be further by its diagonalization
For matrix DT:
DT=VTCTV (5)
Wherein, DTFor CTCharacteristic value according to descending diagonal matrix;V is the orthogonal matrix being made up of characteristic vector.Pass through
Two above step obtains the transformation matrix of MNF:
TMNF:TMNF=PV (6)
Step 3:The encipherment protection of image spectrum feature
The present invention is directed to the feature of remote sensing images, designs a kind of encryption mechanism, this mechanism is being entered to Characteristics of The Remote Sensing Images
While row encipherment protection, the distance between original characteristic vector can also be retained.Specifically it is assumed that using RnTo represent image
Feature, the basic ideas of characteristics of image encipherment protection are exactly to find a kind of encryption function ε (g):Rn→Rm, so that two width
Characteristic vector f of remote sensing images and g, dε(ε (f), ε (g)) ≈ c*d (f, g), wherein dε() and d () are apart from degree
Amount method, and c is constant scale value.The present invention characteristics of image is encrypted from the randomized method of bit-planes, with
When retain characteristics of image between similitude.
Step 3.1:Bit-planes randomization
The most significant bit plane (MSB, Most Significant Bit) of image can reflect the important information of image.?
The thought of important bitplane (MSB) and least significant bit plane (LSB, Least Significant Bit) have begun to for
Multi-media signal is processed.Apart from little characteristic vector, similar distance is also had on MSB plane.
Using this characteristic, can while keeping Hammerstein model, the highest k bit plane of characteristic vector be carried out adding
Close.Give the ciphering process of j-th bit plane of all characteristic vectors in figure 4.Each of composition bit plane is first
The bit sequence random with one does xor operation, thus protecting the information of each bit plane.Again the bit plane obtaining is entered
Row random scrambling is obtaining the bit-planes of encryption.
It is necessary first to carry out the extraction of image bit-plane before extracting important bitplane.So-called image bit-plane is exactly
Each of image pixel value is decomposed into binary value, and the plane that all positions with identical weights are constituted calls position
Plane. for 256 grades of gray level images, each pixel accounts for a byte, i.e. 8 binary digits, by from a high position to low level
It is arranged as b7b6b5b4b3b2blb0, then the b0 position of all pixels just constitutes the 0th bit plane, b1 position just constitutes the 1st position
Plane etc., the rest may be inferred, and this 256 grades of gray level images include 8 bit planes altogether.
Different information bits are different to the contribution of digital picture, the main contents of high bit plane reflection original image,
And the original energy entrained by relatively low bit plane is few, that is, ignore relatively low bit plane to original image quality not too big shadow
Ring.In addition, being known by digital picture correlation theory, after digital picture suffers from attacked by noise (illumination, sharpening, fuzzy etc.), it is low
Bit plane information change is larger, and important bitplane information has almost no change that is to say, that noise is concentrated mainly on compared with low level
Plane.Therefore, extract important plane information to portray and to represent that picture material not only can preferably retain original image information, and
And noise jamming can be prevented effectively from.
The present invention is to extract the bit plane of image feature vector.If f (i) is the value of ith feature vector, special for image
The Bit Plane Decomposition levied may be defined as
Wherein, p=0,1,2,3,4,5,6,7 expression bit plane (p=0 represents bit plane 0, and p=1 represents bit plane 1, according to
This analogizes).
The encryption flow of image feature vector proposed by the present invention is as follows:
Spectra feature extraction is carried out to image, then the span of characteristic value is normalized to [- 1,1], then by feature
Merge into 111 dimensional feature vectors.
In order to carry out bit-planes randomization, after first feature plus 1, it is multiplied by 100 again, now characteristic value obtains span
For [0,200].
Due to 200(10)=11001000(2), characteristic vector can be converted into 8 bit planes, retain three (7-5 of highest
Position), by the 3rd and the 4th bit plane respectively with key XOR, then binary features are switched back to the decimal system, now encrypt
Characteristic value span be [0,223] (223(10)=11011111(2)), finally subtract 1 again divided by 100, characteristics of image encryption
Flow chart such as accompanying drawing 5.
Step 3.2:The similarity measurement of characteristics of image
The present invention calculates similarity using the Hamming distance between feature, is added after characteristics of image is encrypted
Characteristic vector ε (f) after close=[f1,...,fn].The distance between two encrypted features vector ε (f) and ε (g) are by them each
The weighted sum of the Hamming distance between bit plane is calculating:
Wherein, w (j) is assigned to the weight of bit plane, to reflect their different importance.Setting w (j) herein=
2-j, as the weights of each.Due on the corresponding bit plane of two characteristic vectors, using identical permutations and phase
Same exclusive model, to retain their Hamming distance, obtains
Result shows, the characteristic distance after encryption and initial range slightly error.Encrypted feature apart from dε() omits
More than original L1 distance.Error producing cause is:Some vectorial values are likely to be of the L1 distance of very little, but after encryption away from
From dε() is relatively large.For example, 8=(1000)2With 7=(0111)2L1 distance be 1, and dε(8,7)=15.But
It is that such situation seldom occurs, and when weighing with security phase, bit plane randomization only will result only in retrieval precision
Slightly decrease, flow chart such as accompanying drawing 6.
Step 4:Spectral signature similitude is mated
By to curve of spectrum parametrization and extract associated eigenvalue, and then the difference relatively between the different curves of spectrum,
Line retrieval is entered to characteristics of image according to the size sequence of similarity, thus realizing the retrieval of remote sensing images.The spectrum of the present invention
Join process and use spectral information divergence matching method.
Step 4.1:Spectral information divergence is mated
Spectral information divergence coupling is to be determined by calculating the spectral information divergence between target optical spectrum and sample spectrum
Therebetween similitude.It regards each pixel as a stochastic variable, defines its probability using spectrum histogram and divides
Cloth, the computing formula of spectral information divergence is:
SID (t, r)=D (tPr)+D (rPt) (10)
Wherein,
D (tPr) and D (rPt) represent respectively r with regard to the comentropy (K-L distance) of t and t with regard to r comentropy (L-K away from
From);piAnd qiRepresent the probability distribution after target optical spectrum and sample spectrum normalization respectively.
Step 4.2:Relevant feedback
Due to there is very big difference between image low-level feature and high-level semantic, thus by a certain feature extraction with
And the retrieval result obtaining after characteristic vector coupling may tend not to meet user's request.Then, the present invention is based on light
Relevance Feedback is introduced in the image indexing system of spectrum signature.Its target is to interact with the actual of inquiry system from user
Learnt in journey, the actual queries finding and catching user are intended to, with the query strategy of this update the system, obtained real with user
Border demand Query Result as identical as possible.Because relevant feedback can change the query strategy of system in real time, thus for scheming
As searching system increased adaptation function.
During remote Sensing Image Retrieval proposed by the present invention, method is as follows:
First, in the retrieving of the first round, user is first supplied to its sample interested of system, and system is passed through similar
Line retrieval is entered in the calculating of degree, and the image in image library is carried out with similarity size sequence, and the image set in front K position sequence
Present to user as retrieval result;
Then, user can be evaluated to retrieval result and be marked, and points out that in retrieval result, which is and query graph
As related image (positive feedback), which is incoherent image (negative-feedback);
Finally, the relevant information of user's mark is fed back to system as training sample to be learnt, to instruct next round
Retrieval, this completes a relevant feedback process.User can adjust the number of times of relevant feedback according to actual needs, until
To satisfied image set.Flow chart such as accompanying drawing 7.
Finally it should be noted that:Above example only not limits technical side described in the invention in order to the present invention to be described
Case;Therefore, although this specification with reference to above-mentioned example to present invention has been detailed description, this area common
It will be appreciated by the skilled person that still the present invention can be modified or equivalent;And all without departing from invention spirit and
The technical scheme of scope and its improvement, it all should be covered in the middle of scope of the presently claimed invention.
Claims (1)
1. a kind of remote sensing images safe retrieving method based on spectral signature protection is it is characterised in that comprise the steps:
Step 1:Preprocessing of remote sensing images
First deblocking is carried out to remote sensing images, the target image of input is divided into from left to right, from top to bottom regular, do not weigh
Folded and equal-sized subimage block;
Step 2:Remote-sensing image spectrum feature extraction
Data Dimensionality Reduction is carried out using the method separating conversion based on minimal noise, extracts the spectral signature of remote sensing images, method is such as
Under:
Step 2.1:Using high-pass filter, noise extraction is carried out to image, obtain noise covariance matrix CN, and by its diagonalization
For matrix DN, method is as follows:
DN=UTCNU
Wherein, U is the orthogonal matrix being made up of characteristic vector;
Step 2.2:Original image is projected to new space, method is as follows:
I=PTCNP
Step 2.3:Dialogue noise data carries out standard principal component transform, and method is as follows:
CT=PTCDP
Step 2.4:Diagonally turn to matrix DT, method is as follows:
DT=VTCTV
Step 2.5:Calculate minimal noise separation transformation matrix, method is as follows:
TMNF:TMNF=PV
Wherein I is unit battle array, and P is transformation matrix, CDCovariance matrix for image X;CTIt is the matrix after P conversion, DTFor
CTCharacteristic value according to descending diagonal matrix, V is the orthogonal matrix being made up of characteristic vector;
Step 3:The encipherment protection of remote-sensing image spectrum feature
Process remote sensing images using the randomized method of bit-planes, remote-sensing image spectrum feature is encrypted, retain original
The distance between characteristic vector;
Step 4:The similitude coupling of spectral signature
Set up index, carry out remote sensing images coupling using the discrete matching method of spectral information, realize the safe retrieval of remote sensing images.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410149032.8A CN103886106B (en) | 2014-04-14 | 2014-04-14 | Remote sensing image safe-retrieval method based on spectral feature protection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410149032.8A CN103886106B (en) | 2014-04-14 | 2014-04-14 | Remote sensing image safe-retrieval method based on spectral feature protection |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103886106A CN103886106A (en) | 2014-06-25 |
CN103886106B true CN103886106B (en) | 2017-02-22 |
Family
ID=50954998
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410149032.8A Active CN103886106B (en) | 2014-04-14 | 2014-04-14 | Remote sensing image safe-retrieval method based on spectral feature protection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103886106B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR3028064B1 (en) * | 2014-11-05 | 2016-11-04 | Morpho | IMPROVED DATA COMPARISON METHOD |
CN104376121B (en) * | 2014-12-04 | 2018-03-27 | 深圳大数点科技有限公司 | System, method and the user terminal that a kind of picture Adaptive matching combination is presented |
CN104850644B (en) * | 2015-05-27 | 2018-05-29 | 清华大学 | High spectrum image search method and search engine based on Lucene |
CN104933155B (en) * | 2015-06-25 | 2018-05-29 | 清华大学 | Curve of spectrum search method and engine based on Lucene |
CN106874836A (en) * | 2016-12-29 | 2017-06-20 | 国网山东省电力公司菏泽供电公司 | A kind of cable connector running rate recognizing method based on Infrared Thermogram |
CN109144993B (en) * | 2017-06-16 | 2021-07-27 | 腾讯科技(深圳)有限公司 | Data query method and device |
CN108681986B (en) * | 2018-04-28 | 2020-03-31 | 吉林大学 | Method and system for accelerating projection transformation of sine projection format remote sensing product |
CN110490904B (en) * | 2019-08-12 | 2022-11-11 | 中国科学院光电技术研究所 | Weak and small target detection and tracking method |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1452840A (en) * | 2000-05-02 | 2003-10-29 | 通用仪器公司 | Method and apparatus for enabling random access to individual pictures in encrypted video stream |
-
2014
- 2014-04-14 CN CN201410149032.8A patent/CN103886106B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1452840A (en) * | 2000-05-02 | 2003-10-29 | 通用仪器公司 | Method and apparatus for enabling random access to individual pictures in encrypted video stream |
Non-Patent Citations (7)
Title |
---|
"Secure Image Retrieval over Untrusted Cloud Servers";Ayad Ibrahim Adbulsada et al.;《International Journal of Engineering and Advanced Technology》;20131031;第3卷(第1期);第140-147页 * |
"Secure image retrieval through feature protection";Wenjun Lu et al.;《IEEE International Conference on Speech and Signal Processing》;20091231;第2009年卷;第1533-1536页 * |
"一种基于视觉单词的图像检索方法";刁蒙蒙 等;《测控技术》;20121231;第31卷(第5期);第17-20页 * |
"个性化图像检索中用户兴趣模型的构建方法";贺琳 等;《计算机工程与应用》;20091231;第45卷(第31期);第168-171页 * |
"基于最小噪声分离变换的遥感影像融合方法";顾海燕 等;《国土资源遥感》;20070630(第2期);第53-55页 * |
"基于视觉感知的图像检索的研究";张菁 等;《电子学报》;20080331;第36卷(第3期);第494-499页 * |
"高光谱图像特征分析技术的研究与软件开发";张亚南;《中国优秀硕士学位论文全文数据库 信息科技辑》;20100815;第2010年卷(第8期);第1-69页 * |
Also Published As
Publication number | Publication date |
---|---|
CN103886106A (en) | 2014-06-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103886106B (en) | Remote sensing image safe-retrieval method based on spectral feature protection | |
Yu et al. | Attributing fake images to gans: Learning and analyzing gan fingerprints | |
JP7060619B2 (en) | Biometric identification system and method | |
CN104823203A (en) | Biometric template security and key generation | |
Chambino et al. | Multispectral facial recognition: a review | |
Zhao et al. | Spectral–spatial anomaly detection via collaborative representation constraint stacked autoencoders for hyperspectral images | |
Feng et al. | Hyperspectral anomaly detection with total variation regularized low rank tensor decomposition and collaborative representation | |
CN106485742B (en) | A kind of remote sensing images encryption search method based on Arnold chaotic maps | |
CN106650681B (en) | The empty spectrum joint high spectrum image target identification method restored based on low-rank matrix | |
Zhou et al. | Camouflaged target detection based on visible and near infrared polarimetric imagery fusion | |
CN114067188B (en) | Infrared polarized image fusion method for camouflage target | |
Lu et al. | Penalized linear discriminant analysis of hyperspectral imagery for noise removal | |
Jin et al. | Private video foreground extraction through chaotic mapping based encryption in the cloud | |
Zhao et al. | SRA-CEM: An improved CEM target detection algorithm for hyperspectral images based on sub-region analysis | |
Chen et al. | Improving the accuracy of hydrothermal alteration mapping based on image fusion of ASTER and Sentinel-2A data: a case study of Pulang Cu deposit, Southwest China | |
Kaushik et al. | Analysis of spatial features in CBIR system | |
Li et al. | Perceptual hashing for color images | |
CN107832793A (en) | The sorting technique and system of a kind of high spectrum image | |
CN105096234B (en) | A kind of high spectrum image encryption method based on hybrid domain | |
Chen et al. | Hyperspectral target detection based on tensor sparse representation | |
Li et al. | Improving hyperspectral subpixel target detection using hybrid detection space | |
Baek | The fake fingerprint detection system using a novel color distribution | |
Shang et al. | Comparison of bare soil extraction methods in black soil zone for AHSI/GF-5 remote sensing data | |
Zhou et al. | Camouflaged target separation by spectral-polarimetric imagery fusion with shearlet transform and clustering segmentation | |
Shrivastava et al. | Optimization of image retrieval by using HSV color space, Zernike moment & DWT technique |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220819 Address after: 100004 1-14-2107, floor 2, No. 136, Xiwai street, Xicheng District, Beijing Patentee after: Shuanxin (Beijing) Technology Co.,Ltd. Address before: 100124 No. 100 Chaoyang District Ping Tian Park, Beijing Patentee before: Beijing University of Technology |