CN110728635B - Contrast enhancement method for dark and weak target - Google Patents
Contrast enhancement method for dark and weak target Download PDFInfo
- Publication number
- CN110728635B CN110728635B CN201910850365.6A CN201910850365A CN110728635B CN 110728635 B CN110728635 B CN 110728635B CN 201910850365 A CN201910850365 A CN 201910850365A CN 110728635 B CN110728635 B CN 110728635B
- Authority
- CN
- China
- Prior art keywords
- local
- feature
- image
- dark
- pixel
- 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
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000009825 accumulation Methods 0.000 claims abstract description 17
- 238000011176 pooling Methods 0.000 claims abstract description 7
- 238000010801 machine learning Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000003384 imaging method Methods 0.000 claims description 4
- 230000004044 response Effects 0.000 claims description 4
- 238000012706 support-vector machine Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 abstract description 4
- 239000006185 dispersion Substances 0.000 abstract description 2
- 239000013598 vector Substances 0.000 abstract 4
- 238000001514 detection method Methods 0.000 description 14
- 230000000694 effects Effects 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000003331 infrared imaging Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
Images
Classifications
-
- G06T5/94—
-
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a contrast enhancement method of a dark and weak target. Firstly, each frame of gray level image continuously output by an image sensor is obtained, local features of each pixel are extracted, and local feature vectors of each pixel are constructed. And (3) completing calculation of all pixels to obtain a local feature vector diagram of the image, and carrying out pooling operation on the local feature vector diagram. And carrying out local similarity calculation on the feature vector diagrams of the front frame and the rear frame of the sequence image, carrying out local feature probability accumulation according to the similarity, and outputting an enhanced image after each continuous processing of N frames. According to the method, the dark and weak targets are subjected to multidimensional expression, target probability dispersion and matching accumulation through local features, so that randomly distributed space-time noise is reduced, the contrast of the dark and weak targets in an image is improved, and the discovery capability of the dark and weak targets is remarkably improved.
Description
Technical field:
the invention belongs to the technical field of infrared image processing, and relates to a contrast enhancement method of a dark and weak target, which is particularly suitable for image preprocessing enhancement in the process of detecting a target with low signal to noise ratio of an infrared image.
The background technology is as follows:
because the infrared imaging system has the characteristics of long acting distance, high imaging precision, passive imaging and the like, the infrared imaging system is widely applied to detection and tracking of various targets. Low signal to noise ratio dim target detection has been a problem in the infrared image processing field. Because the infrared detection system generally has a longer acting distance, the energy of the target reaching the detection system is weak, and the target often occupies only one or a few pixels on the image plane, and the target detection system has no fixed geometric form or texture information and brings great difficulty to the target detection. Meanwhile, due to the existence of background clutter and detection noise of an infrared image, a target point is submerged in the image noise, and interference is brought to target detection.
At present, two main methods for detecting infrared small targets exist: (1) The image is preprocessed by filtering and the like, and target detection is performed after the background clutter noise is suppressed, so that the aim of suppressing false alarms is fulfilled; (2) Performing target detection after improving contrast by performing multi-frame energy accumulation; (3) And carrying out prediction association on the motion trail of the target of the image sequence so as to realize detection of a small target under low signal-to-noise ratio. (1) The main purpose of the method (3) is to reduce the false alarm rate, and the method does not have a good improvement effect on the detection of a dark and weak target; (2) Good effect on the targets with slower movement speed and poor effect on the dim targets with faster movement speed. It is therefore necessary to develop contrast enhancement methods for the detection of dim targets, improving the signal-to-noise ratio to facilitate the detection of dim targets.
The invention comprises the following steps:
in order to overcome the defects of the prior art, the invention provides a contrast enhancement method for a dark and weak target, which utilizes local feature extraction to carry out multidimensional expression on each pixel, realizes probability dispersion and matching accumulation of the target through feature pooling, reduces randomly distributed space-time noise, improves the contrast of the dark and weak target in an image, and improves the discovery capability of the dark and weak target.
The above object of the present invention is achieved by the following technical solutions:
1. a method for contrast enhancement of a dim target, characterized by: the method comprises the following steps:
(1) Designing a feature extractor and a feature scoring device according to the data or priori information of the dark and weak targets and the clutter background; the data of the dark and weak targets and the clutter background are images shot by a sensor with the same imaging wave band and close resolution; the shooting background type is consistent with the actual type; the prior information is a spatial distribution model of the target and the background and typical information which can characterize the target: intensity, size, energy concentration, shape, speed, background intensity of the target; the feature extractor can be designed manually by using priori knowledge, and can also be designed by using data in an unsupervised machine learning manner, wherein the machine learning method mainly comprises sparse coding, a support vector machine and a deep neural network; the feature scoring device can be designed manually by using priori knowledge, and can also be designed by using unsupervised machine learning through data, wherein the machine learning method mainly comprises sparse coding, a support vector machine and a deep neural network;
(2) Acquiring an output sequence image of an image sensor, wherein the image sensor is an area array staring type image sensor, the response spectrum is not limited, and the output frame frequency is not less than 30 frames per second;
(3) Extracting local features of each pixel in the image by using a feature extractor; the local feature of each pixel is that a W1 & W1 window neighborhood taking each pixel as a center is subjected to slice extraction, and the local feature in the slice is extracted, wherein W1 is the window width and is set manually;
(4) Pooling the local features to obtain a local feature map; the local feature pooling means that the feature of the pixel with the highest local score is selected in each W2-W2 window neighborhood to be used as the output of the neighborhood block, wherein W2 is the window width and is set manually;
(5) Carrying out neighborhood similarity matching on the local feature map of the current frame and the local feature accumulation map of the previous frame, and calculating probability accumulation for updating local features; the neighborhood similarity matching of the previous and the next frames means that the FAI of the characteristic accumulation graph of the previous frame is calculated n-1 Local feature map FI with current frame n The similarity degree of the local characteristics of the pixels in the neighborhood of the same position W3, wherein W3 is a matching window range, the matching window range is set by manpower, and the similarity degree is evaluated as follows:
where n is the current frame number, x 1 、y 1 To accumulate pixel positions of the map, x 2 、y 2 M is the number of frames accumulated currently and is the position of the local feature map of the current frame; performing feature accumulation on pixels most similar to the local features to complete a feature accumulation map FAI n Updating, wherein a, b is an accumulation coefficient, a+b=1;
FAI n =a·FAI n-1 +b·FI n
(6) After the local features of N frames of images are accumulated, feature scoring output is carried out, and image enhancement output is completed, wherein N is the accumulated frame number and can be set in advance; wherein enhancing the image refers to outputting the feature score for each pixel of the local feature map as an image response.
Compared with the prior art, the invention has the beneficial effects that
1. And the target is characterized by local features, and the time space noise is suppressed by the accumulation of similar pixel features of the sequence images, so that the contrast enhancement of the dark and weak target is realized.
2. The image matching operation amount is reduced through local pooling, the operation efficiency of image enhancement and the adaptability of a moving target are improved, the hardware implementation is convenient, and the real-time enhancement is realized.
Drawings
FIG. 1 is a block diagram of an implementation flow of the present invention;
FIG. 2 is an original image of the present invention;
FIG. 3 is an enhanced image of the present invention;
FIG. 4 is a graph showing the enhancement of the present invention for different signal-to-noise targets.
Detailed Description
The following detailed description of the technical solutions according to embodiments of the present invention will be given with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Manually selecting local features, and selecting local maximum values, local average values and local energy concentration;
the manual selection scoring device is a weighted accumulator, three characteristic values are weighted and accumulated, and weight coefficients are respectively [0.3,0.4,0.3];
image resolution 256×256, w1=9, w2=5, w3=3, a=0.8, b=0.2, n=8; the original image is shown in fig. 2, the result after enhancement according to the method flow is shown in fig. 3, and the contrast gain curve of the method is drawn by enhancing dark and weak targets with different signal to noise ratios as shown in fig. 4.
Claims (1)
1. A method for contrast enhancement of a dim target, comprising the steps of:
(1) Designing a feature extractor and a feature scoring device according to the data or priori information of the dark and weak targets and the clutter background; the data of the dark and weak targets and the clutter background are images shot by a sensor with the same imaging wave band and close resolution; the shooting background type is consistent with the actual type; the prior information is a spatial distribution model of the target and the background and typical information which can characterize the target: intensity, size, energy concentration, shape, speed, background intensity of the target; the feature extractor utilizes priori knowledge to carry out manual design or carries out unsupervised machine learning design through data, and the machine learning method comprises sparse coding, a support vector machine and a deep neural network; the feature scoring device utilizes priori knowledge to carry out manual design or carries out unsupervised machine learning design through data, and the machine learning method comprises sparse coding, a support vector machine and a deep neural network;
(2) Acquiring an output sequence image of an image sensor, wherein the image sensor is an area array staring type image sensor, the response spectrum is not limited, and the output frame frequency is not less than 30 frames per second;
(3) Extracting local features of each pixel in the image by using a feature extractor; the local feature of each pixel is that a W1 & W1 window neighborhood taking each pixel as a center is subjected to slice extraction, and the local feature in the slice is extracted, wherein W1 is the window width and is set manually;
(4) Pooling the local features to obtain a local feature map; the local feature pooling means that the feature of the pixel with the highest local score is selected in each W2-W2 window neighborhood to be used as the output of the neighborhood, wherein W2 is the window width and is set manually;
(5) Carrying out neighborhood similarity matching on the local feature map of the current frame and the local feature accumulation map of the previous frame, and calculating probability accumulation for updating local features; the neighborhood similarity matching of the previous and the next frames means that the FAI of the characteristic accumulation graph of the previous frame is calculated n-1 Local feature map FI with current frame n The similarity degree of the local characteristics of the pixels in the neighborhood of the same position W3, wherein W3 is a matching window range, the matching window range is set by manpower, and the similarity degree is evaluated as follows:
where n is the current frame number, x 1 、y 1 To accumulate pixel positions of the map, x 2 、y 2 M is the number of frames accumulated currently and is the position of the local feature map of the current frame; performing feature accumulation on pixels most similar to the local features to complete a feature accumulation map FAI n Updating, wherein a, b is an accumulation coefficient, a+b=1;
FAI n =a·FAI n-1 +b·FI n ;
(6) After the local features of N frames of images are accumulated, feature scoring output is carried out, and image enhancement output is completed, wherein N is the accumulated frame number and can be set in advance; wherein image enhancement refers to outputting the feature score of each pixel of the local feature map as an image response.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910850365.6A CN110728635B (en) | 2019-09-10 | 2019-09-10 | Contrast enhancement method for dark and weak target |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910850365.6A CN110728635B (en) | 2019-09-10 | 2019-09-10 | Contrast enhancement method for dark and weak target |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110728635A CN110728635A (en) | 2020-01-24 |
CN110728635B true CN110728635B (en) | 2023-07-07 |
Family
ID=69218047
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910850365.6A Active CN110728635B (en) | 2019-09-10 | 2019-09-10 | Contrast enhancement method for dark and weak target |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110728635B (en) |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4999711A (en) * | 1988-07-01 | 1991-03-12 | U.S. Philips Corp. | Digital method of modifying an image contrast so as to retain imaging of small objects |
CN101567087A (en) * | 2009-05-25 | 2009-10-28 | 北京航空航天大学 | Method for detecting and tracking small and weak target of infrared sequence image under complex sky background |
CN101930072A (en) * | 2010-07-28 | 2010-12-29 | 重庆大学 | Multi-feature fusion based infrared small dim moving target track starting method |
CN106204476A (en) * | 2016-06-27 | 2016-12-07 | 中国矿业大学 | A kind of low-luminance color image enchancing method |
CN106709426A (en) * | 2016-11-29 | 2017-05-24 | 上海航天测控通信研究所 | Ship target detection method based on infrared remote sensing image |
CN107481210A (en) * | 2017-08-03 | 2017-12-15 | 北京长峰科威光电技术有限公司 | The infrared image enhancing method of local selective mapping based on details |
CN108257155A (en) * | 2018-01-17 | 2018-07-06 | 中国科学院光电技术研究所 | A kind of extension target tenacious tracking point extracting method based on part and Global-Coupling |
CN109767439A (en) * | 2019-01-10 | 2019-05-17 | 中国科学院上海技术物理研究所 | A kind of multiple dimensioned difference of self-adapting window and the object detection method of bilateral filtering |
CN110163235A (en) * | 2018-10-11 | 2019-08-23 | 腾讯科技(深圳)有限公司 | Training, image enchancing method, device and the storage medium of image enhancement model |
-
2019
- 2019-09-10 CN CN201910850365.6A patent/CN110728635B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4999711A (en) * | 1988-07-01 | 1991-03-12 | U.S. Philips Corp. | Digital method of modifying an image contrast so as to retain imaging of small objects |
CN101567087A (en) * | 2009-05-25 | 2009-10-28 | 北京航空航天大学 | Method for detecting and tracking small and weak target of infrared sequence image under complex sky background |
CN101930072A (en) * | 2010-07-28 | 2010-12-29 | 重庆大学 | Multi-feature fusion based infrared small dim moving target track starting method |
CN106204476A (en) * | 2016-06-27 | 2016-12-07 | 中国矿业大学 | A kind of low-luminance color image enchancing method |
CN106709426A (en) * | 2016-11-29 | 2017-05-24 | 上海航天测控通信研究所 | Ship target detection method based on infrared remote sensing image |
CN107481210A (en) * | 2017-08-03 | 2017-12-15 | 北京长峰科威光电技术有限公司 | The infrared image enhancing method of local selective mapping based on details |
CN108257155A (en) * | 2018-01-17 | 2018-07-06 | 中国科学院光电技术研究所 | A kind of extension target tenacious tracking point extracting method based on part and Global-Coupling |
CN110163235A (en) * | 2018-10-11 | 2019-08-23 | 腾讯科技(深圳)有限公司 | Training, image enchancing method, device and the storage medium of image enhancement model |
CN109767439A (en) * | 2019-01-10 | 2019-05-17 | 中国科学院上海技术物理研究所 | A kind of multiple dimensioned difference of self-adapting window and the object detection method of bilateral filtering |
Non-Patent Citations (4)
Title |
---|
J.L.Wang.Fusion of infrared and visible light images based on region feature similarity.《Optics and PRecision Engineering》.2012,263-271页. * |
Taobei Xue et al.A neighboring structure reconstructed matching algorithm based on LARK feature.《Infrared Physics & Technology》.2015,第73卷8-18页. * |
祁伟.基于仿生视觉计算模型的红外图像理解.《中国博士学位论文全文数据库信息科技辑》.2017,第2018年卷(第07期),I138-64. * |
肖宁等.多特征差异决策耦合Top-Hat变换的红外目标检测.《光电工程》 .2016,第43卷(第12期),110-118页. * |
Also Published As
Publication number | Publication date |
---|---|
CN110728635A (en) | 2020-01-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109753903B (en) | Unmanned aerial vehicle detection method based on deep learning | |
CN109767439B (en) | Target detection method for multi-scale difference and bilateral filtering of self-adaptive window | |
CN103533214B (en) | Video real-time denoising method based on kalman filtering and bilateral filtering | |
CN110490904B (en) | Weak and small target detection and tracking method | |
CN104504652A (en) | Image denoising method capable of quickly and effectively retaining edge and directional characteristics | |
CN105913404A (en) | Low-illumination imaging method based on frame accumulation | |
CN106886747B (en) | It is a kind of based on extension wavelet transformation complex background under Ship Detection | |
CN111709888B (en) | Aerial image defogging method based on improved generation countermeasure network | |
CN112686304A (en) | Target detection method and device based on attention mechanism and multi-scale feature fusion and storage medium | |
CN111915558B (en) | Pin state detection method for high-voltage transmission line | |
CN111707998B (en) | Sea surface floating small target detection method based on connected region characteristics | |
CN112070717A (en) | Power transmission line icing thickness detection method based on image processing | |
CN104899842B (en) | The adaptive extreme value median filter method of sequence for remote line-structured light image | |
Long et al. | Underwater forward-looking sonar images target detection via speckle reduction and scene prior | |
CN108508425B (en) | Method for detecting foreground target based on neighborhood characteristics under radar near-earth background noise | |
CN106780545A (en) | A kind of Weak target energy accumulation Enhancement Method of combination time-space domain | |
CN110728635B (en) | Contrast enhancement method for dark and weak target | |
CN107748885B (en) | Method for recognizing fuzzy character | |
CN109285148B (en) | Infrared weak and small target detection method based on heavily weighted low rank and enhanced sparsity | |
CN112764005A (en) | Low signal-to-noise ratio echo data reconstruction method for Gm-APD laser radar combined with morphological filtering | |
CN109410137B (en) | Method for detecting dim and weak target | |
CN115601301B (en) | Fish phenotype characteristic measurement method, system, electronic equipment and storage medium | |
CN110047041A (en) | A kind of empty-frequency-domain combined Traffic Surveillance Video rain removing method | |
CN115393406A (en) | Image registration method based on twin convolution network | |
CN110211124B (en) | Infrared imaging frozen lake detection method based on MobileNet V2 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | 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 |