CN112329764A - Infrared dim target detection method based on TV-L1 model - Google Patents

Infrared dim target detection method based on TV-L1 model Download PDF

Info

Publication number
CN112329764A
CN112329764A CN202011036761.4A CN202011036761A CN112329764A CN 112329764 A CN112329764 A CN 112329764A CN 202011036761 A CN202011036761 A CN 202011036761A CN 112329764 A CN112329764 A CN 112329764A
Authority
CN
China
Prior art keywords
target
image
weak
small
infrared
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011036761.4A
Other languages
Chinese (zh)
Inventor
卢德勇
曹东
刘林岩
杨阳
王海波
赵杨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Computational Aerodynamics Institute of China Aerodynamics Research and Development Center
Original Assignee
Computational Aerodynamics Institute of China Aerodynamics Research and Development Center
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Computational Aerodynamics Institute of China Aerodynamics Research and Development Center filed Critical Computational Aerodynamics Institute of China Aerodynamics Research and Development Center
Priority to CN202011036761.4A priority Critical patent/CN112329764A/en
Publication of CN112329764A publication Critical patent/CN112329764A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/30—Noise filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an infrared small target detection method based on a TV-L1 model, which converts the small target detection problem into an optimization problem, utilizes an alternating direction method to iterate and minimize a target function containing a total variation component and an L1 norm component, effectively separates a background image and a target image, can effectively inhibit background noise and clutter, improves the detection rate of small targets, reduces the false alarm rate caused by interference of edges, noise and the like, greatly reduces the calculation time consumption of the detection method, and improves the real-time property of target detection.

Description

Infrared dim target detection method based on TV-L1 model
Technical Field
The invention belongs to the technical field of image processing and target detection, and particularly relates to a full-variational-based infrared small and weak target detection method.
Background
An infrared detection system, as a passive and passive detection technology, has the advantages of good concealment, less interference from electrons, high reliability and the like, is widely concerned, and has been applied to a plurality of fields, such as infrared search and tracking, accurate guidance, earth observation, monitoring security and the like. With the continuous improvement of application requirements, particularly urgent needs in military aspects such as infrared reconnaissance, early warning and guidance, targets need to be found as far as possible and as early as possible, and therefore, the infrared small and weak target detection technology is always a hot spot and a front edge in the field of target detection.
Generally, a detection algorithm based on a single frame image and a detection algorithm based on a multi-frame image can be simply classified according to the number of images of a target detection algorithm. According to the sequence of the utilized spatial information and the utilized time information, the multi-frame image target detection algorithm can be divided into a track-before-detect algorithm (DBT) and a track-before-detect algorithm (TBD). The strategy adopted by the DBT is that single-frame detection is carried out firstly and then multi-frame confirmation is carried out, preprocessing such as background clutter suppression and weak and small target enhancement is carried out on a single-frame image firstly, then threshold values are set on each frame of image to carry out target detection, a plurality of suspected targets are obtained, and then the targets are confirmed according to priori knowledge of target motion rules and a gray level distribution form. The TBD predicts all possible target motion tracks according to the motion characteristics of the weak and small target motion speed, the motion direction and the like, then calculates the posterior probability of each track according to the gray characteristic of the target, the target pixel size and the target energy change characteristic, and obtains the detection result and the target track after multi-frame accumulation. The three-dimensional matched filter method is a representative algorithm of TBD. In summary, the TBD algorithm essentially exchanges the track information of the target with the accumulation of the number of image frames, so as to obtain better detection and tracking performance, and thus the computation and storage amount are large, and it is difficult to implement in a real-time system. The DBT algorithm has the advantages of clear logic, simple implementation and easy hardware modularization implementation, but has the problems of poor anti-interference capability, insufficient reliability and low detection rate. The multi-frame image target detection algorithm can improve the detection performance, but the algorithm is high in complexity, long in calculation time consumption and difficult to meet the real-time requirement, partial prior assumptions are difficult to meet, and prior information is difficult to obtain, so that the applicability of the multi-frame image target detection method is limited.
Compared with a target detection method based on a multi-frame image, the weak and small target detection method based on a single-frame image has the following characteristics: (1) the algorithm is relatively simple, the calculation time is short, and the real-time requirement is easily met; (2) the method can adapt to scenes with complex backgrounds and fast changes; (3) the method for detecting the small and weak targets of the single-frame image is not only the basis of a plurality of multi-frame image-based target detection algorithms, but also the key and the core of the whole target detection algorithm. Therefore, the research based on the single-frame image target detection method is always an important research content of the infrared weak and small target detection algorithm. The main technical approaches comprise: a filtering based method, a Human Visual System (HVS) based method, a low rank matrix recovery based method, etc.
The filtering-based method mainly detects small and weak infrared targets by suppressing the background, and can be divided into a processing method based on a spatial domain and a processing method based on a transform domain according to different processing angles. The spatial domain filtering method comprises spatial domain high-pass filtering, median filtering, MaxMean, MaxMedian, Robinson Guard filtering, bilateral filtering, TopHat filtering and the like. The method based on transform domain filtering includes classical frequency domain high-pass filtering, wavelet filtering, Hilbert-Huang transform filtering and the like. The detection algorithm based on filtering, especially the algorithm based on the spatial domain, generally speaking, has low complexity and small calculation amount, but due to the fact that priori knowledge is difficult to obtain, for example, a template or a structural element is difficult to determine, the application range of the algorithm is narrow, the stability of the algorithm is insufficient, and the false alarm rate is high; the algorithm based on the transform domain increases the processing and operation on the transform domain, increases the calculation amount, and is often difficult to meet the real-time requirement.
The infrared small and weak target detection method based on the human visual system mainly constructs a saliency map capable of highlighting small and weak targets through local differences of the targets and the background, and further achieves target detection. The key is how to effectively capture the salient region. Researchers have detected small weak targets by evaluating the difference or discontinuity between a small weak target region and its neighboring regions. Chen et al propose the concept of Local Contrast Measures (LCM) that compute the dissimilarity between the current location and its neighbors and target the most prominent points of the corresponding target saliency map. Deng et al introduced a Weighted Local Differential Metric (WLDM) to represent the discontinuity, combining multi-scale local contrast with local entropy. The detection algorithm based on the human visual system is an important direction for intelligent research, but at present, a saliency map is constructed by mostly only using information in the aspect of local contrast, so that the requirement of high reliability is difficult to meet. In order to improve the detection rate of the algorithm and reduce the false alarm rate, multi-dimensional information and deep information need to be extracted to construct a saliency map, such as spatial information and transform domain information, such as entropy, correlation, directional relation, etc., but this increases the complexity of the algorithm.
Due to good data dimension exploration capability, the low-rank matrix theory is widely concerned in the fields of computer vision, machine learning, artificial intelligence and the like. Gao et al propose a small and weak target detection method for a single-frame infrared image, convert the small and weak target detection problem into an optimization problem of separation and recovery of a low-rank matrix and a sparse matrix, and then solve the problem by using a Robust Principal Component Analysis (RPCA) method and an accelerated near-end gradient method (APG). Then, many researchers carry out further research and provide a series of infrared small and weak target detection algorithms based on low-rank matrix recovery. Some improve the infrared block image IPI model, put forward the infrared block tensor IPT model, and some improve the objective function for solving the low-rank matrix recovery optimization problem. Tensor et al propose replacing the nuclear norm with the Partial Sum Tensor Nuclear Norm (PSTNN) in a weighted infrared block tensor (RIPT) model, and then converting the target background separation problem into a robust low-rank tensor recovery problem. The method based on low rank matrix recovery can adapt to the situation that the signal-to-noise ratio is low, and can have higher reliability under a more complex background, however, some problems still exist. For example, in the aspect of modeling, a block matrix obtained by an infrared image block model may not meet a low-rank condition, so that the convergence rate of an optimization algorithm for recovering the low-rank matrix is low; in the aspect of an algorithm, the image block model obviously increases the matrix scale to be processed, the algorithm for solving the optimization problem is high in complexity and long in calculation time, the real-time requirement is difficult to meet, and a related fast algorithm needs to be developed to improve the convergence speed.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a full-variational-based infrared weak and small target detection method, effectively solves the problems of low detection rate and high false alarm rate caused by factors such as noise, clutter and edges in the prior art, has less calculation time and easily meets the real-time requirement.
The purpose of the invention is realized by the following technical scheme:
an infrared small and weak target detection method based on a TV-L1 model converts an infrared small and weak target detection problem into an objective function optimization problem containing a total variation component item of a background image and an L1 norm item of a target image:
Figure BDA0002705293590000031
in the formula: d, B, T and N are the original infrared image, background image, target image and noise image, | | gTVRepresenting the fully-variant norm, | | g | | non-conducting phosphor1Represents the norm L1, λ is a constant greater than zero;
in the process of solving the optimization problem, an alternative direction method ADM is adopted to obtain the solved target image TkAnd performing threshold segmentation on the target image from the target image TkAnd obtaining the final weak target of the segmentation.
According to a preferred embodiment, the optimization problem solving process comprises:
the method comprises the following steps: inputting an original infrared image D of a current frame, and selecting values of parameters lambda and mu;
step two: assigning an initial value T to the target image0＝0，k＝0；
Step three: according to the TV-L1 model, an objective function of an optimization problem is established:
Figure BDA0002705293590000032
in the formula: | g | calculation of luminanceFDenotes the Frobenius norm, μ being a penalty parameter.
Step four: and (3) utilizing an alternating direction method to iteratively minimize the objective function P (B, T) to obtain two sub-optimization problems:
Figure BDA0002705293590000033
Figure BDA0002705293590000034
step five: alternately solving the two sub-optimization problems until the iterative method converges to obtain the solved background image BkAnd a target image Tk；
Step six: from the target image T by a threshold segmentation methodkAnd obtaining the final weak target of the segmentation.
According to a preferred embodiment, in the fourth step, the specific solving method of the two sub-optimization problems is as follows:
Bk+1is solved for D-TkCarrying out a TV denoising process and solving an approximate solution;
Tk+1solving by a soft threshold operator:
Figure BDA0002705293590000041
wherein
Figure BDA0002705293590000042
According to a preferred embodiment, in step four, Bk+1The approximation solution can be found using a gradient projection algorithm or a chambole algorithm or a Moreau approximation operator.
According to a preferred embodiment, in the sixth step, the threshold calculation formula is:
Tth＝mean+θσ
in the formula, mean is the mean value of the target image, sigma is the variance of the target image, theta is a constant, and theta ranges from 3 to 8.
According to a preferred embodiment, the gray value of a pixel in the target image is greater than TthThe hour is indicated as the target.
The main scheme and the further selection schemes can be freely combined to form a plurality of schemes which are all adopted and claimed by the invention; in the invention, the selection (each non-conflict selection) and other selections can be freely combined. The skilled person in the art can understand that there are many combinations, which are all the technical solutions to be protected by the present invention, according to the prior art and the common general knowledge after understanding the scheme of the present invention, and the technical solutions are not exhaustive herein.
The invention has the beneficial effects that: the method of the invention converts the detection problem of the weak and small targets into an optimization problem, utilizes an alternating direction method to iterate and minimize a target function containing a total variation item and an L1 norm item, effectively separates out a background image and a target image, can effectively inhibit background noise and clutter, improves the detection rate of the weak and small targets, reduces the false alarm rate caused by interference of edges, noise and the like, greatly reduces the calculation time consumption of the detection method, and improves the real-time property of target detection.
Drawings
FIG. 1 is a flow chart of an optimization problem solving process in the object detection method of the present invention.
Fig. 2 is a schematic diagram of an infrared image mathematical model adopted by the invention.
Fig. 3 is an infrared image containing a small target used in an embodiment of the present invention.
Fig. 4 is a target image obtained by solving the original infrared image of fig. 3 in the embodiment of the present invention.
FIG. 5 is a diagram of a comparison of infrared image targets detected by the method of the present invention with other methods.
FIG. 6 is a diagram of a comparison of infrared image targets detected by the method of the present invention with other methods.
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.
It should be noted that, in order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described below, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments.
Thus, the following detailed description of the embodiments of the present invention is not intended to limit the scope of the invention as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate orientations and positional relationships that are conventionally used in the products of the present invention, and are used merely for convenience in describing the present invention and for simplicity in description, but do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed in a particular orientation, and be operated, and therefore, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," "third," and the like are used solely to distinguish one from another and are not to be construed as indicating or implying relative importance.
Furthermore, the terms "horizontal", "vertical", "overhang" and the like do not imply that the components are required to be absolutely horizontal or overhang, but may be slightly inclined. For example, "horizontal" merely means that the direction is more horizontal than "vertical" and does not mean that the structure must be perfectly horizontal, but may be slightly inclined.
In the description of the present invention, it should also be noted that, unless otherwise explicitly specified or limited, the terms "disposed," "mounted," "connected," and "connected" are to be construed broadly and may, for example, be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
In addition, it should be noted that, in the present invention, if the specific structures, connection relationships, position relationships, power source relationships, and the like are not written in particular, the structures, connection relationships, position relationships, power source relationships, and the like related to the present invention can be known by those skilled in the art without creative work on the basis of the prior art.
Example 1:
referring to fig. 1, the invention discloses an infrared small and weak target detection method based on a TV-L1 model. Converting the infrared weak and small target detection problem into an objective function optimization problem containing a total variation component term of a background image and an L1 norm term of a target image:
Figure BDA0002705293590000061
in the formula: d, B, T and N are respectively an original infrared image, a background image, a target image and a noise image, | g | | countTVRepresenting the fully-variant norm, | | g | | non-conducting phosphor1Represents the norm L1, λ is a constant greater than zero;
in the optimization problem solving process, an original infrared image is input, initial values are given to parameters, two sub-optimization problems are solved alternately, a background image, a target image and a parameter k value are updated until a convergence condition is met, the target image is obtained, then a weak and small target is segmented from the target image through a threshold segmentation method, and weak and small target information is output.
Preferably, the optimization problem solving process includes:
the method comprises the following steps: the original infrared image D of the current frame is input and the values of the parameters λ and μ are selected.
Step two: assigning an initial value T to the target image0＝0，k＝0。
Step three: according to the TV-L1 model, an objective function of an optimization problem is established:
Figure BDA0002705293590000062
in the formula: | g | calculation of luminanceFDenotes the Frobenius norm, μ being a penalty parameter.
Step four: and (3) utilizing an alternating direction method to iteratively minimize the objective function P (B, T) to obtain two sub-optimization problems:
Figure BDA0002705293590000063
Figure BDA0002705293590000064
preferably, the specific solving method of the two sub-optimization problems is as follows: b isk+1Is solved for D-TkCarrying out a TV denoising process and solving an approximate solution;
Tk+1solving by a soft threshold operator:
Figure BDA0002705293590000065
wherein
Figure BDA0002705293590000066
Further, the step fourIn (B)k+1The approximation solution can be found using a gradient projection algorithm or a chambole algorithm or a Moreau approximation operator.
Step five: alternately solving the two sub-optimization problems until the iterative method converges to obtain the solved background image BkAnd a target image Tk。
Step six: from the target image T by a threshold segmentation methodkAnd obtaining the final weak target of the segmentation.
Preferably, in the sixth step, the threshold calculation formula is:
Tth＝mean+θσ
in the formula, mean is the mean value of the target image, sigma is the variance of the target image, theta is a constant, and theta ranges from 3 to 8.
Preferably, when the gray value of the pixel in the target image is greater than TthThe hour is indicated as the target.
The method of the invention converts the detection problem of the weak and small targets into an optimization problem, utilizes an alternating direction method to iterate and minimize a target function containing a total variation item and an L1 norm item, effectively separates out a background image and a target image, can effectively inhibit background noise and clutter, improves the detection rate of the weak and small targets, reduces the false alarm rate caused by interference of edges, noise and the like, greatly reduces the calculation time consumption of the detection method, and improves the real-time property of target detection.
Referring to fig. 2 to 6, the effect analysis is performed according to the attached drawings:
fig. 2 is a schematic diagram of an infrared image mathematical model adopted by the invention, wherein each part of the diagram is a target image and a background image obtained by using the method of the invention. The lower left corner of each sub-graph is a partial enlarged view of a small target, which shows that the mathematical model has good applicability. Fig. 3 is an infrared image containing a small object. Fig. 4 is a partially enlarged view of a small target detected by the method of the present invention, and the lower left corner illustrates that the present invention can detect the small target. Fig. 5 and 6 are diagrams comparing infrared image weak targets detected by the method of the present invention with other methods, wherein the first row is an original image in a real infrared image sequence. The second line is the target detection result of the TopHat filtering method. The third row is the target detection result of the IPI model. The fourth row is the target detection result of the PSTNN model. The fifth row is the target detection result of the method of the invention. The detection results of the small targets are marked with white boxes. In the weak and small target detection of the 10 images, the TopHat method has 4 false alarms, and other methods have no false alarm; in the used calculation time, 44.7076 seconds are used in an IPI method, 1.4667 seconds are used in a PSTNN method, the real-time requirement cannot be met, 0.2933 seconds are used in the method, about 30 frames of images are detected every second, and the real-time requirement of weak and small target detection is easily met. The result shows that the method can improve the detection precision, reduce the false alarm rate and meet the real-time requirement.
The foregoing basic embodiments of the invention and their various further alternatives can be freely combined to form multiple embodiments, all of which are contemplated and claimed herein. In the scheme of the invention, each selection example can be combined with any other basic example and selection example at will. Numerous combinations will be known to those skilled in the art.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (6)

1. A method for detecting infrared weak and small targets based on a TV-L1 model is characterized in that,
converting the infrared weak and small target detection problem into an objective function optimization problem containing a total variation component term of a background image and an L1 norm term of a target image:
Figure FDA0002705293580000011
in the formula: d, B, T and N are respectively an original infrared image, a background image, a target image and a noise image, | g | | countTVRepresenting the fully-variant norm, | | g | | non-conducting phosphor1Denotes the L1 norm, λ is oneA constant greater than zero;
in the process of solving the optimization problem, an alternative direction method ADM is adopted to obtain the solved target image TkAnd performing threshold segmentation on the target image from the target image TkAnd obtaining the final weak target of the segmentation.
2. The TV-L1 model-based infrared small dim target detection method according to claim 1, wherein the optimization problem solving process comprises:
the method comprises the following steps: inputting an original infrared image D of a current frame, and selecting values of parameters lambda and mu;
step two: assigning an initial value T to the target image0＝0，k＝0；
Step three: according to the TV-L1 model, an objective function of an optimization problem is established:
Figure FDA0002705293580000012
in the formula: | g | calculation of luminanceFDenotes the Frobenius norm, μ being a penalty parameter.
Step four: and (3) utilizing an alternating direction method to iteratively minimize the objective function P (B, T) to obtain two sub-optimization problems:
Figure FDA0002705293580000013
Figure FDA0002705293580000014
step five: alternately solving the two sub-optimization problems until the iterative method converges to obtain the solved background image BkAnd a target image Tk；
Step six: from the target image T by a threshold segmentation methodkAnd obtaining the final weak target of the segmentation.
3. The method for detecting infrared weak and small targets based on the TV-L1 model as claimed in claim 2, wherein in the fourth step, two sub-optimization problems are solved specifically by:
Bk+1is solved for D-TkCarrying out a TV denoising process and solving an approximate solution;
Tk+1solving by a soft threshold operator:
Figure FDA0002705293580000021
wherein
Figure FDA0002705293580000022
4. The TV-L1 model-based infrared small and weak target detection method as claimed in claim 3, wherein in step four, Bk+1The approximation solution can be found using a gradient projection algorithm or a chambole algorithm or a Moreau approximation operator.
5. The TV-L1 model-based infrared small and weak target detection method as claimed in claim 2, wherein in the sixth step, the threshold calculation formula is:
Tth＝mean+θσ
in the formula, mean is the mean value of the target image, sigma is the variance of the target image, theta is a constant, and theta ranges from 3 to 8.
6. The TV-L1 model-based infrared weak and small object detection method as claimed in claim 5, wherein when the gray value of the pixel in the object image is greater than TthThe hour is indicated as the target.
CN202011036761.4A 2020-09-28 2020-09-28 Infrared dim target detection method based on TV-L1 model Pending CN112329764A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011036761.4A CN112329764A (en) 2020-09-28 2020-09-28 Infrared dim target detection method based on TV-L1 model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011036761.4A CN112329764A (en) 2020-09-28 2020-09-28 Infrared dim target detection method based on TV-L1 model

Publications (1)

Publication Number Publication Date
CN112329764A true CN112329764A (en) 2021-02-05

Family

ID=74303388

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011036761.4A Pending CN112329764A (en) 2020-09-28 2020-09-28 Infrared dim target detection method based on TV-L1 model

Country Status (1)

Country Link
CN (1) CN112329764A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112802020A (en) * 2021-04-06 2021-05-14 中国空气动力研究与发展中心计算空气动力研究所 Infrared dim target detection method based on image inpainting and background estimation
CN113421279A (en) * 2021-07-07 2021-09-21 中国科学院光电技术研究所 Infrared weak and small target detection method based on weighted nuclear norm minimization
CN114166850A (en) * 2021-11-30 2022-03-11 电子科技大学 Light-excited infrared thermal imaging defect detection method based on differential tensor decomposition

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107392095A (en) * 2017-06-15 2017-11-24 杭州红辉电子科技有限公司 A kind of small IR targets detection algorithm based on mask image
CN109325446A (en) * 2018-09-19 2019-02-12 电子科技大学 A kind of method for detecting infrared puniness target based on weighting truncation nuclear norm
CN109584303A (en) * 2018-12-03 2019-04-05 电子科技大学 A kind of method for detecting infrared puniness target based on Lp norm and nuclear norm
CN110135344A (en) * 2019-05-16 2019-08-16 电子科技大学 Based on the method for detecting infrared puniness target for weighting fixed order expression

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107392095A (en) * 2017-06-15 2017-11-24 杭州红辉电子科技有限公司 A kind of small IR targets detection algorithm based on mask image
CN109325446A (en) * 2018-09-19 2019-02-12 电子科技大学 A kind of method for detecting infrared puniness target based on weighting truncation nuclear norm
CN109584303A (en) * 2018-12-03 2019-04-05 电子科技大学 A kind of method for detecting infrared puniness target based on Lp norm and nuclear norm
CN110135344A (en) * 2019-05-16 2019-08-16 电子科技大学 Based on the method for detecting infrared puniness target for weighting fixed order expression

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王忠美等: "基于鲁棒主成分分析的红外图像小目标检测", 《兵工学报》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112802020A (en) * 2021-04-06 2021-05-14 中国空气动力研究与发展中心计算空气动力研究所 Infrared dim target detection method based on image inpainting and background estimation
CN113421279A (en) * 2021-07-07 2021-09-21 中国科学院光电技术研究所 Infrared weak and small target detection method based on weighted nuclear norm minimization
CN113421279B (en) * 2021-07-07 2023-03-31 中国科学院光电技术研究所 Infrared weak and small target detection method based on weighted nuclear norm minimization
CN114166850A (en) * 2021-11-30 2022-03-11 电子科技大学 Light-excited infrared thermal imaging defect detection method based on differential tensor decomposition
CN114166850B (en) * 2021-11-30 2023-06-09 电子科技大学 Light excitation infrared thermal imaging defect detection method based on differential tensor decomposition

Similar Documents

Publication Publication Date Title
CN110472627B (en) End-to-end SAR image recognition method, device and storage medium
CN109740665B (en) Method and system for detecting ship target with occluded image based on expert knowledge constraint
CN108648161B (en) Binocular vision obstacle detection system and method of asymmetric kernel convolution neural network
Wang et al. A region based stereo matching algorithm using cooperative optimization
Kamencay et al. Improved Depth Map Estimation from Stereo Images Based on Hybrid Method.
CN112329764A (en) Infrared dim target detection method based on TV-L1 model
Bu et al. Detection of fabric defects by auto-regressive spectral analysis and support vector data description
CN110782477A (en) Moving target rapid detection method based on sequence image and computer vision system
Chai et al. Stereo matching algorithm based on joint matching cost and adaptive window
CN109559273B (en) Quick splicing method for vehicle bottom images
CN106815576B (en) Target tracking method based on continuous space-time confidence map and semi-supervised extreme learning machine
CN110660065B (en) Infrared fault detection and identification algorithm
CN109949345A (en) The method neural network based for obtaining lack of balance SAR image variation detection figure
CN113205494B (en) Infrared small target detection method and system based on adaptive scale image block weighting difference measurement
CN116863285A (en) Infrared and visible light image fusion method for multiscale generation countermeasure network
CN115861669A (en) Infrared dim target detection method based on clustering idea
CN112348853B (en) Particle filter tracking method based on infrared saliency feature fusion
CN113888428A (en) Infrared dim target detection method and device based on local contrast
Yin Image micro-target recognition method based on multi-layer empirical mode decomposition algorithm
Liu et al. Research on A Binocular Fish Dimension Measurement Method Based on Instance Segmentation and Fish Tracking
CN116524008B (en) Target object matching and spatial position estimation method for security inspection CT intelligent identification
He et al. A high-precision optical flow method based on TV-L1 model and neighborhood patch
CN113658191B (en) Infrared dim target detection method based on local probability hypergraph dissimilarity measure
Immella et al. How to Improve the Detection of Ships from Satellite Images by employing Convolutional Neural Network Using Convolutional Block Attention Modules
Ji et al. Adaptive underwater sonar image target detection

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20210205

RJ01 Rejection of invention patent application after publication