CN105551029B - A kind of multi-spectral remote sensing image Ship Detection - Google Patents
A kind of multi-spectral remote sensing image Ship Detection Download PDFInfo
- Publication number
- CN105551029B CN105551029B CN201510905953.7A CN201510905953A CN105551029B CN 105551029 B CN105551029 B CN 105551029B CN 201510905953 A CN201510905953 A CN 201510905953A CN 105551029 B CN105551029 B CN 105551029B
- Authority
- CN
- China
- Prior art keywords
- spectral
- notable
- indicates
- ship
- value
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10036—Multispectral image; Hyperspectral image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of methods that multi-spectral remote sensing image Ship Target quickly detects, and belong to Remote Sensing Target detection technique field.This method takes full advantage of the information of spectral domain and spatial domain, is processed to the Walsh Hadamard transform domain coefficient of multi-spectral remote sensing image to extract the significant characteristics of spatial domain, is detected for ShipTargets.The present invention can effectively overcome traditional multi-spectral image ship detection method computation complexity high, the disadvantage of parameter setting complexity, it is through a large amount of true Multi-spectral Remote Sensing Datas the experimental results showed that, the present invention has fast and accurately ShipTargets detection effect, while having stronger robustness to spectral coverage noise and mixed and disorderly marine background.In marine fisheries management, sea transport control, marine military monitoring etc. has huge application value.
Description
Technical field
The invention belongs to Remote Sensing Target detection technique fields, and in particular to a kind of multi-spectral remote sensing image Ship Target
The method quickly detected.
Background technique
Multispectral imaging sensor can obtain the spectral information of each spectral coverage tie substance in spectrum dimension, while in space dimension
The spatial information of scene is obtained, is formed and contains the multidimensional data body of abundant geographical environmental information.Due to manually build target with
Natural material background has biggish difference in the spectral characteristic of each spectral coverage, so multispectral image data are examined in ground object target
There is unique advantage in survey, can be used in Automatic Targets task.Especially maritime transportation, marine fisheries management with
And military monitoring etc., the marine vessel detection of multi-spectral remote sensing image more have great significance.
Common multispectral object detection method is normally based on the statistical detection method of spectral information, i.e., by assuming that
Measured value is made of background, target and noise, using the model of method the difference tectonic setting and target of statistics, is recycled false
If examining the testing result for obtaining differentiating target, representative method therein is exactly that the multichannel of the propositions such as Reed is permanent empty
Alert rate (constant false alarm rate, CFAR) method.In the multispectral image of different scenes, spectrum picture
Statistical property is variation, the target of CFAR method be it is adaptive find a detection threshold value, in different spectrum pictures
Constant detection false alarm rate is maintained in object detection task.CFAR method primary disadvantage is that, if the spectrum picture of target is believed
Number similar to surrounding environment in gray level, then necessarily will cause false-alarm, Automatic Targets will be become difficult.
Although multispectral imaging sensor can detect for the man-made target in mixed and disorderly background and provide help, in automatic warship
Two difficulties will be faced in the practical application of ship detection.Firstly, sea wind, presence situations such as ocean current, the tail of ship, oil leak will
Cause seawater radiation or the variation of spectral reflection characteristic.The spectral reflection characteristic on naval vessel is also unstable in traveling, and is difficult to
Estimation.All these situations can all cause sea clutter (sea clutter) and Ship Target in the system of each spectral coverage radiation energy
The overlapping scored on cloth.Secondly, marine monitoring system needs a kind of quick algorithm of target detection, since it needs analysis in real time
With a large amount of multispectral data of processing, this requirement is also challenged for automatic Target Detection.In order to be distinguished from false-alarm
Interesting target out, conventional method usually require the algorithm comprising target identification, which comes generally for real-time system
Say that operation is excessively complicated.In addition, conventional method is generally required to all image-regions in order to which the existence to target judges
It is verified, but content actually of concern usually only accounts for very small part in image.This comprehensive working process both can
It causes to calculate and waste, and aggravated analysis difficulty.
Summary of the invention
It is an object of the invention to propose a kind of multi-spectral remote sensing image naval vessel detection side of view-based access control model conspicuousness mechanism
Method, computation complexity is low, parameter setting is simple, can accurately and effectively detect naval vessel mesh in multi-spectral remote sensing image
Mark.
To reach above-mentioned purpose of the invention, multi-spectral remote sensing image Ship Detection provided by the invention is specific to wrap
Include following steps：
1) l for the multi-spectral remote sensing image that spatial resolution is M × N pixel is tieed up into spectroscopic data and is considered as l width gray level image
Xi, i=1 ..., l carry out Walsh Hadamard transform to every one-dimensional spectrum picture respectively；
2) in Walsh Hadamard transform domain, the Walsh Hadamard transform domain coefficient of every one-dimensional spectrum picture is carried out
Normalization operation sets 1 for all positive value elements, sets -1 for all negative value elements；
3) corresponding Walsh Hadamard inverse transformation is carried out to every one-dimensional normalized Walsh Hadamard transform domain coefficient,
Calculate the spectral coverage notable figure per one-dimensional spectrum picture；
4) the spectral coverage notable figure of all dimension spectrum pictures is summed on Spatial Dimension, is then filtered with Gaussian smoothing
Wave device carries out slight smoothing processing to the result of summation, calculates final synthesis notable figure；
5) go out the detection threshold value of Ship Target according to the mean value computation of final notable figure；
6) the automatic detection to Ship Target is realized using the detection threshold value of Ship Target, obtain the detection knot of binaryzation
Fruit.
Wherein, in the first step, the specific formula for calculation of Walsh Hadamard transform is carried out such as to input multispectral image
Under：
Fi=HXiWT, i=1 ..., l
Wherein, H indicates that M rank hadamard matrix, W indicate N rank hadamard matrix, FiIndicate i-th dimension spectrum picture XiWall
Assorted Hadamard transform domain coefficient matrix.
Meanwhile the spatial resolution of the image of input be M × N, and M and N meet Walsh Hadamard transform sequence it is long
The requirement of degree, if not satisfied, the mode of operation for carrying out piecemeal processing to entire image can be taken, in order to retain small and weak naval vessel
Signal can not carry out down-sampled processing to the original image of input.
Wherein, in second step, Walsh Hadamard transform domain coefficient is normalized the specific formula for calculation of operation
For：
Wherein, | | indicate the symbol that takes absolute value, BiIndicate i-th dimension spectrum picture XiNormalized Walsh Hadamard
Coefficient in transform domain matrix.
Wherein, in the third step, the specific formula for calculation of the spectral coverage notable figure per one-dimensional spectrum picture is as follows：
Si=abs (HTBi), W i=1 ..., l
Wherein, abs () indicates the operation that takes absolute value to each element of input matrix.
Wherein, in the 4th step, the specific formula for calculation of final synthesis notable figure is：
Wherein, G indicates the Gaussian kernel of 2 dimensions, and S indicates finally obtained synthesis notable figure.
The present invention is in the calculating process of the spectral coverage notable figure of every one-dimensional spectrum picture, high-frequency noise (such as sea clutter noise)
Also can be by violent amplification, but these high-frequency noises being amplified are incoherent in each spectral coverage, and each spectral coverage is significant
Ship signaling in figure is relevant.Therefore, each spectral coverage notable figure is added, can allows these in each spectral coverage notable figure
Incoherent high-frequency noise is cancelled out each other, and is mutually inhibited, but also can allow the warship of statistical correlation in each spectral coverage notable figure
Ship signal, which is overlapped mutually, to be further enhanced.Nevertheless, the result after the summation of all spectral coverage notable figuresIn still
Some high-frequency noises not curbed can be so remained, the result that the presence of these noises can be detected to naval vessel brings false-alarm, is
Eliminate these residual high-frequency noises may bring detect false-alarm, the present invention uses a Gaussian kernel with suitable parameters
The result that G is added all spectral coverage notable figuresLow pass smothing filtering is carried out, to allow ship signaling and ambient noise signal to have
There are apparent grey scale contrasts, while ship signaling can also be retained well.
Wherein, in the 5th step, the specific formula for calculation of the detection threshold value of Ship Target is：
Wherein, θ is the detection threshold value of Ship Target, and M and N are the length and width of notable figure, and α is from the more of multiple groups different scenes
The empirical value obtained in spectral image data naval vessel test experience, by many experiments, discovery setting 3≤α≤5 can be obtained
Obtain preferable testing result.
When multispectral image is there are when Ship Target, the Ship Target in synthesis notable figure is highlighted, saliency value
It is bigger, and marine background region compares darker, saliency value is smaller.In this case, the saliency value meeting of Ship Target
Mean value than synthesizing notable figure is much bigger, and also greater than the detection threshold value obtained by notable figure mean value computation, then naval vessel
Target can be come out by accurate detection.When multispectral image does not have Ship Target, would not occur in synthesis notable figure
Particularly pertinent highlighted target.In this case, the maximum value for synthesizing notable figure will not be bigger than synthesizing the average value of notable figure
Very much, and the maximum value of notable figure can also be less than the detection threshold value obtained by synthesizing notable figure mean value computation, so as to avoid
The generation of false-alarm.
Wherein, in step 6, it is to the specific formula for calculation that Ship Target detects automatically：
Wherein, D is binaryzation testing result, and will test image-region judgement of the value equal to 1 is Ship Target, be will test
Image-region judgement of the value equal to 0 is marine background.
Multi-spectral remote sensing image Ship Target Detection method proposed by the invention, is the side of view-based access control model conspicuousness mechanism
Method, this method are obtained notable figure using the Walsh Hadamard transform domain coefficient of multispectral image data and for Ship Targets
Detection.Compared with traditional multispectral object detection method, the present invention is not needed with the probability distribution to Sea background and target
Some Utopian premises assumed as modeling of characteristic；It is independent of priori knowledge, also without the parameter of many complexity
Setting.Simultaneously as Walsh Hadamard transform, which exists, fast implements algorithm, the method in the present invention can quickly calculate defeated
The notable figure of the multispectral data entered can meet the requirement handled in real time in practical application, be multi-spectral remote sensing image naval vessel mesh
Mark detection provides a kind of new effective fast algorithm.
Detailed description of the invention
Fig. 1 is the multi-spectral remote sensing image Ship Detection flow chart of implementation column of the present invention；
Fig. 2 is naval vessel detection example of the present invention to the multi-spectral remote sensing image for having Ship Target；
Wherein：(a)-(f) 6 spectral coverages of the true Landsat7 remotely-sensed data of East China Sea partial region；(g) it calculates
The synthesis notable figure arrived；(h) testing result；
Fig. 3 is naval vessel detection example of the present invention to the multi-spectral remote sensing image of no Ship Target；
Wherein：(a)-(f) 6 spectral coverages of the true Landsat7 remotely-sensed data of East China Sea partial region；(g) it calculates
The synthesis notable figure arrived；(h) testing result；
Specific embodiment
Below by example, the present invention will be further described.It should be noted that the purpose for publicizing and implementing example is to help
It helps and further understands the present invention, but it will be appreciated by those skilled in the art that：The present invention and appended claims are not being departed from
Spirit and scope in, various substitutions and modifications are all possible.Therefore, the present invention should not be limited to interior disclosed in embodiment
Hold, the scope of protection of present invention is subject to the scope defined in the claims.
Fig. 1 is the process flow diagram of multi-spectral remote sensing image Ship Detection of the present invention, including：
Multispectral data is carried out Walsh Hadamard transform per one-dimensional spectrum picture by the first step
The l for the multi-spectral remote sensing image that spatial resolution is M × N pixel is tieed up into spectroscopic data and is considered as l width gray level image Xi,
I=1 ..., l carries out Walsh Hadamard transform to every one-dimensional spectrum picture respectively, and specific formula for calculation is as follows：
Fi=HXiWT, i=1 ..., l
Wherein, H indicates that M rank hadamard matrix, W indicate N rank hadamard matrix, FiIndicate i-th dimension spectrum picture XiWall
Assorted Hadamard transform domain coefficient matrix.
Operation is normalized to the Walsh Hadamard transform domain coefficient of every one-dimensional spectrum picture in second step
In Walsh Hadamard transform domain, the Walsh Hadamard transform domain coefficient of every one-dimensional spectrum picture is returned
One changes operation, i.e., sets 1 for all positive value elements, set -1 for all negative value elements, specific formula for calculation is as follows：
Wherein, | | indicate the symbol that takes absolute value, BiIndicate i-th dimension spectrum picture XiNormalized Walsh Hadamard
Coefficient in transform domain matrix.
Third step does Walsh Hadamard inverse transformation to every one-dimensional normalization coefficient, calculates per one-dimensional spectrum picture
Spectral coverage notable figure
Corresponding Walsh Hadamard inverse transformation, meter are carried out to every one-dimensional normalized Walsh Hadamard transform domain coefficient
The spectral coverage notable figure per one-dimensional spectrum picture is calculated, specific formula for calculation is as follows：
Si=abs (HTBi), W i=1 ..., l
Wherein, abs () indicates the operation that takes absolute value to each element of input matrix.
All spectral coverage notable figures are added to obtain width synthesis notable figure by the 4th step
The spectral coverage notable figure of all dimension spectrum pictures is summed on Spatial Dimension, it is aobvious to calculate final synthesis
Figure is write, specific formula for calculation is as follows：
Wherein, G indicates the Gaussian kernel of 2 dimensions, and S indicates finally obtained synthesis notable figure.In this step, the present invention is used
One is 0.75 (0.5~1) having a size of 3 × 3 (or 5 × 5), standard deviation, and the gauss low frequency filter of rotational symmetry is to addition
ResultCarry out slight smoothing processing.
5th step goes out the detection threshold value of Ship Target according to synthesis notable figure mean value computation
Go out the detection threshold value of Ship Target according to the mean value computation of final notable figure, specific formula for calculation is as follows：
Wherein, θ is the detection threshold value of Ship Target, and M and N are the length and width of notable figure, and α is from the more of multiple groups different scenes
The empirical value obtained in spectral image data naval vessel test experience, by many experiments, discovery setting 3≤α≤5 can be obtained
Preferable testing result is obtained, such as takes the notable figure mean value of α=4 times as detection threshold value, which is having the case where Ship Target
The lower saliency value that can be less than Ship Target in synthesis notable figure, and synthesis notable figure can be greater than in the case where no Ship Target
Maximum saliency value, that is to say, that it not only can accurately detect Ship Target in the scene for having Ship Target, but also can be
Do not have to effectively prevent detection false-alarm in the scene of Ship Target.
6th step realizes the automatic detection to Ship Target using detection threshold value
The automatic detection to Ship Target is realized using the detection threshold value of Ship Target, obtains the testing result of binaryzation,
Specific formula for calculation is as follows：
Wherein, D is binaryzation testing result, and region of the value equal to 1 indicates Ship Target, and region of the value equal to 0 indicates
Marine background.
It is illustrated in figure 2 above-mentioned treatment process and handles the example that a scape has the multi-spectral remote sensing image of Ship Target.Fig. 2
(a)-(f) show the image of 6 spectral coverages of the true Landsat7 remotely-sensed data of East China Sea partial region, wherein including
Several Ship Target signals.In the synthesis notable figure that this method shown in Fig. 2 (g) is calculated, notable figure is full resolution
, Ship Target is enhanced and is come out from complex background saliency.It show what this method was calculated in Fig. 2 (h)
Ship Target Detection is as a result, white area indicates that the Ship Target detected, black region indicate marine background.
Being illustrated in figure 3 one scape of above-mentioned treatment process processing does not have the example of the multi-spectral remote sensing image of Ship Target.Fig. 3
(a)-(f) show the image of 6 spectral coverages of the true Landsat7 remotely-sensed data of East China Sea partial region, without appoint
What Ship Target signal.In the synthesis notable figure that this method shown in Fig. 3 (g) is calculated, due to no any naval vessel mesh
Mark, so the marking area not enhanced especially in notable figure.The naval vessel mesh that this method shown in Fig. 3 (h) is calculated
It marks in testing result, entire area is that black indicates that detection does not cause any false-alarm.
Multi-spectral remote sensing image Ship Detection disclosed by the invention, merely with each dimension spectrogram of multispectral data
The coefficient of the Walsh Hadamard transform of picture calculates the conspicuousness of corresponding scene, the normalization operation that uses to coefficient in transform domain
With biorational, it simulates the process that homogenous characteristics mutually inhibit in human brain primary visual cortex, can obtain complete point
The notable figure of resolution simultaneously comes out Ship Target from marine background saliency, uncorrelated using ambient noise in each spectral coverage notable figure
And ship signaling statistical correlation this characteristic, each spectral coverage notable figure is subjected to summation on Spatial Dimension to effectively weaken and press down
The high-frequency noise in notable figure has been made, can both have Ship Target come the detection threshold value determined using synthesis notable figure mean value
Accurate detection is to Ship Target in scene, and detection false-alarm, side can be effectively avoided in the scene of not Ship Target
Method is simple, efficient, accurate.The present invention achieves the knot for being substantially better than other conventional methods in the test of a large amount of multispectral datas
Fruit.
Although the present invention has been disclosed in the preferred embodiments as above, however, it is not intended to limit the invention.It is any to be familiar with ability
The technical staff in domain, without departing from the scope of the technical proposal of the invention, all using in the methods and techniques of the disclosure above
Appearance makes many possible changes and modifications or equivalent example modified to equivalent change to technical solution of the present invention.Therefore,
Anything that does not depart from the technical scheme of the invention are made to the above embodiment any simple according to the technical essence of the invention
Modification, equivalent variations and modification, all of which are still within the scope of protection of the technical scheme of the invention.
Claims (8)
1. a kind of multi-spectral image ship detection method, which is characterized in that include the following steps：
1) l for the multi-spectral remote sensing image that spatial resolution is M × N pixel is tieed up into spectroscopic data and is considered as l width gray level image Xi, i=
1 ..., l carries out Walsh Hadamard transform to every one-dimensional spectrum picture respectively；
2) in Walsh Hadamard transform domain, normalizing is carried out to the Walsh Hadamard transform domain coefficient of every one-dimensional spectrum picture
Change operation, i.e., sets 1 for all positive value elements, set -1 for all negative value elements；
3) corresponding Walsh Hadamard inverse transformation is carried out to every one-dimensional normalized Walsh Hadamard transform domain coefficient, calculated
Spectral coverage notable figure per one-dimensional spectrum picture out；
4) the spectral coverage notable figure of all dimension spectrum pictures is summed on Spatial Dimension, then uses Gaussian filter
The result of summation is smoothed, final synthesis notable figure is calculated；
5) go out the detection threshold value of Ship Target according to the mean value computation of final notable figure；
6) the automatic detection to Ship Target is realized using the detection threshold value of Ship Target, obtain the testing result of binaryzation.
2. multi-spectral image ship detection method as described in claim 1, which is characterized in that the Walsh hada in step 1)
Hadamard transform specific formula for calculation is as follows：
Fi=HXiWT, i=1 ..., l
Wherein, H indicates that M rank hadamard matrix, W indicate N rank hadamard matrix, FiIndicate i-th dimension spectrum picture XiWalsh breathe out
Up to Hadamard transform domain coefficient matrix.
3. multi-spectral image ship detection method as described in claim 1, which is characterized in that the space of image point in step 1)
Resolution is M × N, and wherein M and N meets the requirement of the sequence length of Walsh Hadamard transform, if not satisfied, taking to whole picture figure
As carrying out piecemeal processing.
4. multi-spectral image ship detection method as described in claim 1, which is characterized in that Walsh hada in step 2)
The specific formula for calculation that operation is normalized in Hadamard transform domain coefficient is：
Wherein, | | indicate the symbol that takes absolute value, Fi(x, y) indicates i-th dimension spectrum picture XiWalsh Hadamard transform domain system
Matrix number FiMiddle coordinate is the coefficient of (x, y), Bi(x, y) indicates i-th dimension spectrum picture XiNormalized Walsh Hadamard become
Change domain coefficient matrix BiMiddle coordinate is the coefficient of (x, y).
5. multi-spectral image ship detection method as described in claim 1, which is characterized in that every one-dimensional spectrum in step 3)
The specific formula for calculation of the spectral coverage notable figure of image is as follows：
Si=abs (HTBi), W i=1 ..., l
Wherein, abs () indicates the operation that takes absolute value to each element of input matrix；H indicates M rank hadamard matrix, W table
Show N rank hadamard matrix；BiIndicate i-th dimension spectrum picture XiNormalized Walsh Hadamard transform domain coefficient matrix.
6. multi-spectral image ship detection method as described in claim 1, which is characterized in that final synthesis is aobvious in step 4)
The specific formula for calculation of work figure is：
Wherein, G indicates the Gaussian kernel of 2 dimensions, S1,S2,...,SlIndicate the spectral coverage notable figure per one-dimensional spectrum picture；S indicates final
Obtained synthesis notable figure.
7. multi-spectral image ship detection method as described in claim 1, which is characterized in that Ship Target in step 5)
The specific formula for calculation of detection threshold value is：
Wherein, S (x, y) indicates that coordinate is the significant value coefficient of (x, y) in finally obtained synthesis notable figure S, and θ is Ship Target
Detection threshold value, M and N are the length and width of notable figure, and α is the multispectral image data naval vessel test experience from multiple groups different scenes
One empirical value of middle acquisition, 3≤α≤5.
8. multi-spectral image ship detection method as described in claim 1, which is characterized in that in step 6) certainly to Ship Target
Moving the specific formula for calculation detected is：
Wherein, S (x, y) indicates that coordinate is the significant value coefficient of (x, y) in finally obtained synthesis notable figure S, and D (x, y) is two
Coordinate is the testing result coefficient of (x, y) in value testing result matrix D, and region of the value equal to 1 indicates Ship Target, and value is equal to
0 region indicates marine background.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510905953.7A CN105551029B (en) | 2015-12-09 | 2015-12-09 | A kind of multi-spectral remote sensing image Ship Detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510905953.7A CN105551029B (en) | 2015-12-09 | 2015-12-09 | A kind of multi-spectral remote sensing image Ship Detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105551029A CN105551029A (en) | 2016-05-04 |
| CN105551029B true CN105551029B (en) | 2018-11-20 |
Family
ID=55830205
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510905953.7A Expired - Fee Related CN105551029B (en) | 2015-12-09 | 2015-12-09 | A kind of multi-spectral remote sensing image Ship Detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105551029B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106886760B (en) * | 2017-01-24 | 2019-08-16 | 北京理工大学 | A kind of EO-1 hyperion Ship Detection combined based on empty spectrum information |
| CN109934801A (en) * | 2019-01-25 | 2019-06-25 | 淮阴师范学院 | A kind of image Focus field emission array implementation method based on piecemeal Hadamard transform |
| CN109977892B (en) * | 2019-03-31 | 2020-11-10 | 西安电子科技大学 | Ship detection method based on local saliency features and CNN-SVM |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4236767B2 (en) * | 1999-06-25 | 2009-03-11 | 三菱スペース・ソフトウエア株式会社 | Method for detecting movement information of moving object based on satellite SAR image |
| CN102096824A (en) * | 2011-02-18 | 2011-06-15 | 复旦大学 | Multi-spectral image ship detection method based on selective visual attention mechanism |
| CN103729848A (en) * | 2013-12-28 | 2014-04-16 | 北京工业大学 | Hyperspectral remote sensing image small target detection method based on spectrum saliency |
-
2015
- 2015-12-09 CN CN201510905953.7A patent/CN105551029B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4236767B2 (en) * | 1999-06-25 | 2009-03-11 | 三菱スペース・ソフトウエア株式会社 | Method for detecting movement information of moving object based on satellite SAR image |
| CN102096824A (en) * | 2011-02-18 | 2011-06-15 | 复旦大学 | Multi-spectral image ship detection method based on selective visual attention mechanism |
| CN103729848A (en) * | 2013-12-28 | 2014-04-16 | 北京工业大学 | Hyperspectral remote sensing image small target detection method based on spectrum saliency |
Non-Patent Citations (3)
| Title |
|---|
| Bottom-Up Visual Saliency Using Binary Spectrum of Walsh-Hadamard Transform;Ying Yu et al;《International Conference on Neural Information Processing》;20141231;第36-37页第2.2节 * |
| SALIENCY-BASED SHIP DETECTION IN SAR IMAGES;Ying Yu et al;《The Institution of Engineering & Technology》;20151014;第3页第3.1-3.2节，第4页第3.3节 * |
| 选择性视觉注意机制下的多光谱图像舰船检测;丁正虎等;《计算机辅助设计与图形学学报》;20110331;第23卷(第3期);第419-425页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105551029A (en) | 2016-05-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Qi et al. | A robust directional saliency-based method for infrared small-target detection under various complex backgrounds | |
| Xia et al. | Infrared small target detection based on multiscale local contrast measure using local energy factor | |
| Dong et al. | Robust infrared maritime target detection based on visual attention and spatiotemporal filtering | |
| CN106384344B (en) | A kind of remote sensing image surface vessel target detection and extracting method | |
| CN105354541B (en) | The SAR image object detection method of view-based access control model attention model and constant false alarm rate | |
| Nasiri et al. | Infrared small target enhancement based on variance difference | |
| CN109427055B (en) | Remote sensing image sea surface ship detection method based on visual attention mechanism and information entropy | |
| CN111027497B (en) | Weak and small target rapid detection method based on high-resolution optical remote sensing image | |
| Hou et al. | SAR image ship detection based on visual attention model | |
| CN102096824A (en) | Multi-spectral image ship detection method based on selective visual attention mechanism | |
| CN105184804B (en) | Small targets detection in sea clutter method based on Airborne IR camera Aerial Images | |
| Song et al. | Automatic ship detection for optical satellite images based on visual attention model and LBP | |
| CN105551029B (en) | A kind of multi-spectral remote sensing image Ship Detection | |
| Corbane et al. | Fully automated procedure for ship detection using optical satellite imagery | |
| Cai et al. | Automatic circular oil tank detection in high-resolution optical image based on visual saliency and Hough transform | |
| CN106291550B (en) | The polarization SAR Ship Detection of core is returned based on local scattering mechanism difference | |
| Ma et al. | A sea-sky line detection method based on line segment detector and Hough transform | |
| Zhu et al. | Saliency‐Based Diver Target Detection and Localization Method | |
| Yu et al. | Automated ship detection from optical remote sensing images | |
| Hwang et al. | An efficient ship detection method for KOMPSAT-5 synthetic aperture radar imagery based on adaptive filtering approach | |
| Zhu et al. | Unsupervised object-based differencing for land-cover change detection | |
| Picard et al. | Seafloor description in sonar images using the monogenic signal and the intrinsic dimensionality | |
| Haigang et al. | A novel ship detection method for large-scale optical satellite images based on visual LBP feature and visual attention model | |
| Huang et al. | Infrared small target detection with directional difference of Gaussian filter | |
| Qu et al. | Feature-level fusion of dual-band infrared images based on gradient pyramid decomposition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181120 Termination date: 20211209 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |