CN107256559B - Method for complex background suppression - Google Patents
Method for complex background suppression Download PDFInfo
- Publication number
- CN107256559B CN107256559B CN201710406241.XA CN201710406241A CN107256559B CN 107256559 B CN107256559 B CN 107256559B CN 201710406241 A CN201710406241 A CN 201710406241A CN 107256559 B CN107256559 B CN 107256559B
- Authority
- CN
- China
- Prior art keywords
- image
- detection device
- infrared image
- interlaced
- formula
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G06T2207/10048—Infrared 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
- G06T2207/20004—Adaptive image processing
-
- 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
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
-
- 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
-
- 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
- G06T2207/20048—Transform domain processing
- G06T2207/20056—Discrete and fast Fourier transform, [DFT, FFT]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
Abstract
The method comprises the steps that a detection device receives an input infrared image I (x, y) to be detected and a pyramid layer number L, median filtering preprocessing is utilized to remove salt-pepper noise, a processed infrared image f (x, y) is obtained, the row number of the processed infrared image f (x, y) is determined, the highest pyramid layer number lmax is calculated according to the row number, whether the pyramid layer number L exceeds the lmax is judged, if yes, complex background suppression processing is stopped, otherwise, frequency domain Gaussian low-pass filtering is conducted on the processed infrared image f (x, y) to obtain a filtered image g (x, y), interlaced alternate downsampling is conducted on the obtained filtered image g (x, y), the downsampling times are counted in an accumulated mode, if the counted value reaches N-1, the obtained downsampled image is recorded as h (x, y), image h (x, y) is subjected to image expansion through a bicubic interpolation algorithm, a complex background estimation image is obtained, the preprocessed infrared image and the complex estimation image are subjected to subtraction, and a high adaptive detection result is obtained.
Description
Technical Field
The invention relates to the technical field of remote sensing, in particular to a method for inhibiting a complex background.
Background
In recent years, the background suppression technology is an indispensable link for automatic detection of infrared small targets, is mainly applied to the fields of national defense, security and the like, is one of important technologies related to national safety, and has a self-evident importance. Background suppression technologies at home and abroad are mainly classified into two categories, one is a traditional background suppression method based on a filter, and the other is a background suppression method based on background classification.
The conventional background suppression method based on the filter is mainly based on that the local radiation intensity of the infrared small target object is larger than that of the background clutter relative to a gentle background, so that the local gray scale of the infrared small target object appears as a prominent bright spot in an image and is changed violently. This type of method exploits this property to propose gradient-based background suppression algorithms, such as: the well-known top-Hat transform, the Gaussian high-pass filtering algorithm and the like are easy to be interfered by salt and pepper noise and are very sensitive to the salt and pepper noise because the gradient information is used as a judgment standard. Since this approach assumes that the background is flat, false alarms tend to occur when there are severe changes in the background, degrading performance, for example: strong interference such as cloud cluster and cloud fog appears in the sky background.
Compared with the traditional method, the background suppression method based on background classification considers that the background is gently changed, and the background is changed and random, so that the background needs to be estimated firstly, namely the background is simulated. The main modes at the present stage are as follows: the background suppression algorithm based on complexity classification, the background suppression algorithm based on optical flow estimation and the like are all a mode of firstly extracting some local, global and time-varying information in an image through an original image, then simulating the background of the original image by utilizing the information, and finally subtracting the background from the original image to realize the suppression function of the background. Compared with the traditional method, the method can solve the problem that the background contains strong interference, but the actual application is not wide due to the deviation between the simulated background and the background in the actual image and the possible weakening of target energy
Disclosure of Invention
The invention aims to solve the technical problems that the detection accuracy is reduced and the false alarm rate is increased due to the complex background in the infrared weak and small target detection, and aims to solve the technical problems.
In order to solve the above technical problem and achieve the above object, the present invention adopts a method of complex background suppression, comprising:
the method comprises the steps of receiving an input infrared image I (x, y) to be detected and a pyramid layer number L, preprocessing the input infrared image I (x, y) to be detected and the pyramid layer number L by utilizing median filtering to remove salt and pepper noise to obtain a preprocessed infrared image f (x, y), determining the row number and column number of the preprocessed infrared image f (x, y), calculating the highest pyramid layer number 1max according to the row number, judging whether the pyramid layer number L exceeds 1max, stopping complex background suppression processing if the pyramid layer number exceeds 1max, performing frequency domain Gaussian low-pass filtering on the preprocessed infrared image f (x, y) by the detection device to obtain a filtered image g (x, y), performing interlaced alternate column downsampling on the obtained filtered image g (x, y) by the detection device, simultaneously performing cumulative counting on downsampling times, recording the downsampled image obtained at the moment as h (x, y) if the cumulative counting number reaches N-1, performing image expansion on the downsampled image h (x, y) by the detection device by utilizing a bicubic interpolation algorithm to obtain a complex estimated background image, subtracting the background image and obtaining a complex estimated image by the detection device.
Optionally, the detection device performs frequency domain gaussian low-pass filtering on the preprocessed infrared image f (x, y) to obtain a filtered image g (x, y), which specifically includes: the detection device carries out two-dimensional discrete Fourier transform on the preprocessed infrared image f (x, y) to convert the infrared image f (x, y) into a frequency domain, and the formula is as follows:
obtaining a Gaussian low-pass filter template through a Gaussian low-pass filter formula, wherein the formula is as follows:
and (3) carrying out phase-point multiplication on the frequency domain image F (u, v) and the filter template H (u, v) to realize filtering operation, and obtaining a filtered image G (u, v) by the formula:
G(u，v)＝∑u∑vF(u，v)×H(u，v)
and finally, only converting the filtered image G (u, v) into a space domain, namely performing inverse two-dimensional discrete Fourier transform on the filtered image G (u, v), wherein the formula is as follows:
resulting in a filtered image g (x, y).
Optionally, the detecting device performs interlaced-to-interlaced downsampling on the obtained filtered image g (x, y), specifically: if the number of rows and columns is odd, then one row or column is reduced accordingly on the filtered image and the reduced filtered image g (x, y) is then down-sampled interlaced.
Optionally, the detecting device performs interlaced-to-interlaced downsampling on the obtained filtered image g (x, y), specifically: said detecting means using formulas
The filtered image g (x, y) is down-sampled interlaced.
Optionally, the detecting device performs image expansion on the down-sampled image h (x, y) by using a bicubic interpolation algorithm, so as to obtain a complex background estimation image, specifically: the detection device utilizes a bicubic interpolation formula to improve the resolution of the down-sampled image to be the same as that of the preprocessed image, wherein the bicubic interpolation formula is as follows:
the technical scheme of the invention has the following beneficial effects:
in the above scheme, the background information is directly extracted from the original image by using the image golden sub-tower algorithm, and the type of the background is not limited, so that the algorithm itself can adapt to various complex background conditions, such as: and weak and small targets and the like under the sky background containing cloud layers have higher adaptability. Can be applied to the fields of military, security protection and the like. The invention uses the multi-resolution technology of the image pyramid algorithm, can greatly reserve the target energy, reduce the loss of the target energy and facilitate the subsequent operation compared with the traditional method.
Drawings
Fig. 1 is a schematic flow chart of a complex background suppression method according to the present invention.
Fig. 2 is a diagram illustrating the effect of the complex background suppression method according to the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
The invention provides a method for inhibiting a complex background based on an image pyramid, which has high detection rate and low calculation complexity, aims at solving the problems of low detection accuracy and high false alarm rate caused by the complex background in the detection of infrared weak and small targets, is applied to a detection device and comprises the following steps as shown in figure 1:
step 1, a detection device receives an input infrared image I (x, y) to be detected, and performs pretreatment by using median filtering to remove salt and pepper noise, so as to obtain a pretreated infrared image f (x, y);
step 2, the detection device receives the input pyramid layer number L;
and 3, determining the row and column number of the preprocessed infrared image f (x, y) by the detection device, calculating the highest pyramid layer number 1max according to the row and column number, judging whether the input pyramid layer number L exceeds 1max, if so, terminating the complex background suppression processing, and if not, continuing to execute the step 4.
The detection device calculates the number of rows and columns of the infrared image by using the preprocessed infrared image f (x, y), and the number of rows and columns are respectively recorded as M and N, and because the down-sampling method adopts an interlaced-alternate-column down-sampling method, that is, the image after each down-sampling is 1/4 of the original image, the highest layer number 1max of the pyramid is limited by M and N, and the formula is as follows:
Mmax＝log2M
Nmax＝log2N
1max＝min{Mmax，Nmax} (1)
and 4, performing frequency domain Gaussian low-pass filtering on the preprocessed infrared image f (x, y) by the detection device to obtain a filtered image g (x, y), namely a filtering result, wherein the cut-off frequency usually takes an empirical value.
The detection device carries out two-dimensional discrete Fourier transform on the preprocessed infrared image f (x, y) to convert the infrared image f (x, y) into a frequency domain, and the formula is as follows:
according to the preprocessed infrared image f (x, y), the size information of the infrared image f (x, y) can be known, and corresponding row and column numbers are obtained and are respectively marked as M and N. Determining the cut-off frequency D by combining the actual situation0The gaussian low-pass filter template can be obtained by the gaussian low-pass filter formula, which is:
then, multiplying the frequency domain image F (u, v) obtained before by the filter template H (u, v) to complete the filtering operation, which is expressed by the following formula:
G(u，v)＝∑u∑vF(u，v)×H(u，v) (4)
and finally, only converting the filtered image G (u, v) into a space domain, namely performing inverse two-dimensional discrete Fourier transform on the image G (u, v), wherein the formula is as follows:
the obtained filtering image g (x, y) is the result of frequency domain gaussian low-pass filtering.
And 5, performing interlaced alternate column downsampling on the obtained filtering image g (x, y) by the detection device, if the number of the rows and the columns is an odd number, correspondingly reducing one row or one column on the filtering image, then performing interlaced alternate column downsampling on the reduced filtering image g (x, y), and meanwhile, performing cumulative counting on the downsampling times.
The detection device first acquires lines of the filtered image g (x, y)Number and column number, and respectively denoted as M1And N1. Then according to the number of lines M1And number of columns N1If the number of the images is an even number, corresponding processing is carried out, if the number of the images is the even number, the down-sampling operation is directly carried out, otherwise, the corresponding last row (or column) of the image g (x, y) is abandoned, and then the down-sampling operation is carried out. The formula is as follows:
the down-sampling mode is interlaced alternate column down-sampling, and the formula is as follows:
and 6, the detection device repeatedly executes the steps 4 and 5 until the count reaches N-1, and the loop is exited. And recording the down-sampled image obtained at the moment as h (x, y);
7, the detection device performs image expansion on the image h (x, y) obtained in the step 6 by using a bicubic interpolation algorithm so as to obtain a complex background estimation image;
the detection device utilizes a bicubic interpolation algorithm to improve the resolution of the image with the smaller size obtained in the step 6 to be the same as the resolution of the image obtained in the step 1, and because the bicubic interpolation algorithm carries out cubic interpolation on the gray values of 16 points around the sampling point, the influence of the gray values of 4 directly adjacent points is considered, and the influence of the change rate of the gray values of all adjacent points is considered, and the formula is as follows:
and 8, carrying out subtraction operation on the preprocessed infrared image obtained in the step 1 and the complex background estimation image obtained in the step 7 to obtain a detection result. As shown in fig. 2, which is a diagram illustrating the effect of the method for complex background suppression implemented by the method of the present invention.
The invention directly extracts background information from the original image by applying the image golden sub-tower algorithm without limiting the types of the backgrounds, so that the algorithm can adapt to various complex background conditions, such as: and weak and small targets and the like under the sky background containing cloud layers have higher adaptability. Can be applied to the fields of military, security protection and the like. The invention uses the multi-resolution technology of the image pyramid algorithm, can greatly reserve the target energy, reduce the loss of the target energy and facilitate the subsequent operation compared with the traditional method.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (5)
1. A method of complex background suppression, comprising:
the detection device receives an input infrared image I (x, y) to be detected and a pyramid layer number L, performs pretreatment by using median filtering to remove salt and pepper noise, obtains a pretreated infrared image f (x, y),
the detection device determines the row and column number of the preprocessed infrared image f (x, y), calculates a pyramid maximum layer number lmax according to the row and column number, judges whether the pyramid layer number L exceeds the pyramid maximum layer number lmax, and terminates complex background suppression processing if the pyramid layer number L exceeds the pyramid maximum layer number lmax;
the detection device performs interlaced alternate downsampling on the obtained filtering image g (x, y), and counts the downsampling times in an accumulated way; if the accumulated count value reaches N-1, the obtained downsampled image is recorded as h (x, y), the detection device calculates the number of rows and columns of the infrared image by using the preprocessed infrared image f (x, y), and the rows and columns are respectively recorded as M and N, because the downsampling method adopts an interlaced and spaced downsampling method, namely the image downsampled each time is 1/4 of the original image, the maximum number lmax of layers of the pyramid is limited by M and N, and the formula is as follows:
the detection device performs frequency domain Gaussian low-pass filtering on the preprocessed infrared image f (x, y) to obtain a filtered image g (x, y), namely a filtering result, wherein the cut-off frequency here usually takes an empirical value;
the detection device performs image expansion on the down-sampled image h (x, y) by using a bicubic interpolation algorithm so as to obtain a complex background estimation image;
and the detection device subtracts the preprocessed infrared image and the complex background estimation image to obtain a detection result.
2. The method according to claim 1, wherein the detection device performs a frequency domain gaussian low pass filtering on the preprocessed infrared image f (x, y) to obtain a filtered image g (x, y), specifically:
the detection device carries out two-dimensional discrete Fourier transform on the preprocessed infrared image f (x, y) to convert the infrared image f (x, y) into a frequency domain, and the formula is as follows:
obtaining a Gaussian low-pass filter template through a Gaussian low-pass filter formula, wherein the formula is as follows:
and (3) carrying out phase-point multiplication on the frequency domain image F (u, v) and the filter template H (u, v) to realize filtering operation, and obtaining a filtered image G (u, v) by the formula:
G(u,v)＝∑u∑vF(u,v)×H(u,v)
and finally, only converting the filtered image G (u, v) into a space domain, namely performing inverse two-dimensional discrete Fourier transform on the filtered image G (u, v), wherein the formula is as follows:
the resulting filtered image g (x, y).
3. The method according to claim 1, characterized in that the detection means down-sample the obtained filtered image g (x, y) interlaced by interlaced intervals, in particular:
if the number of rows or columns is odd, a corresponding row or column reduction is made on the filtered image, and the reduced filtered image g (x, y) is then down-sampled interlaced.
4. The method according to claim 1, characterized in that the detection means down-sample the obtained filtered image g (x, y) interlaced by interlaced intervals, in particular:
said detecting means using formulas
The filtered image g (x, y) is down-sampled interlaced.
5. The method according to claim 1, wherein the detection device performs image expansion on the down-sampled image h (x, y) by using a bicubic interpolation algorithm to obtain a complex background estimation image, specifically:
the detection device utilizes a bicubic interpolation formula to improve the resolution of the down-sampled image to be the same as that of the preprocessed image, wherein the bicubic interpolation formula is
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710406241.XA CN107256559B (en) | 2017-06-01 | 2017-06-01 | Method for complex background suppression |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710406241.XA CN107256559B (en) | 2017-06-01 | 2017-06-01 | Method for complex background suppression |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107256559A CN107256559A (en) | 2017-10-17 |
| CN107256559B true CN107256559B (en) | 2020-07-14 |
Family
ID=60022943
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710406241.XA Active CN107256559B (en) | 2017-06-01 | 2017-06-01 | Method for complex background suppression |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107256559B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102222322A (en) * | 2011-06-02 | 2011-10-19 | 西安电子科技大学 | Multiscale non-local mean-based method for inhibiting infrared image backgrounds |
| CN102324021A (en) * | 2011-09-05 | 2012-01-18 | 电子科技大学 | Infrared dim-small target detection method based on shear wave conversion |
| CN105809147A (en) * | 2016-03-29 | 2016-07-27 | 北京环境特性研究所 | Earth atmosphere infrared background inhibition method based on Markov autoregression model |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7551334B2 (en) * | 2005-07-20 | 2009-06-23 | Xerox Corporation | Background suppression method and apparatus |
| CN100461218C (en) * | 2007-03-29 | 2009-02-11 | 杭州电子科技大学 | Method for enhancing medical image with multi-scale self-adaptive contrast change |
| US7936923B2 (en) * | 2007-08-31 | 2011-05-03 | Seiko Epson Corporation | Image background suppression |
| CN102419972B (en) * | 2011-11-28 | 2013-02-06 | 西安交通大学 | Method of detecting and identifying sound signals |
| CN102945548A (en) * | 2012-11-20 | 2013-02-27 | 成都晶石石油科技有限公司 | Directional pyramid filtering-based image processing method and device |
| CN103177458B (en) * | 2013-04-17 | 2015-11-25 | 北京师范大学 | A kind of visible remote sensing image region of interest area detecting method based on frequency-domain analysis |
| CN103645467B (en) * | 2013-11-11 | 2016-04-06 | 北京环境特性研究所 | The method and system of target detection in ocean clutter cancellation and sea clutter background |
| CN103632348B (en) * | 2013-11-29 | 2016-05-11 | 中国人民解放军国防科学技术大学 | A kind of multiple dimensioned geometric representation method of nonlinear images |
| CN106384344B (en) * | 2016-08-30 | 2019-02-22 | 中国科学院长春光学精密机械与物理研究所 | A kind of remote sensing image surface vessel target detection and extracting method |
-
2017
- 2017-06-01 CN CN201710406241.XA patent/CN107256559B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102222322A (en) * | 2011-06-02 | 2011-10-19 | 西安电子科技大学 | Multiscale non-local mean-based method for inhibiting infrared image backgrounds |
| CN102324021A (en) * | 2011-09-05 | 2012-01-18 | 电子科技大学 | Infrared dim-small target detection method based on shear wave conversion |
| CN105809147A (en) * | 2016-03-29 | 2016-07-27 | 北京环境特性研究所 | Earth atmosphere infrared background inhibition method based on Markov autoregression model |
Non-Patent Citations (2)
| Title |
|---|
| Adaptive detection for infrared small target undersea-skycomplexbackground;L. Yang等;《ELECTRONICS LETTERS》;20040831;第40卷(第17期);第1-2页 * |
| 基于改进的剪切波变换和引导滤波的红外弱小目标背景抑制方法;荣生辉 等;《光子学报》;20150228;第44卷(第2期);第1-6页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107256559A (en) | 2017-10-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103761731A (en) | Small infrared aerial target detection method based on non-downsampling contourlet transformation | |
| CN104657945A (en) | Infrared small target detection method for multi-scale spatio-temporal union filtering under complex background | |
| CN106469313A (en) | A kind of detection method of small target of caliber adaptive space-time domain filtering | |
| CN110580709A (en) | Target detection method based on ViBe and three-frame differential fusion | |
| CN107886526A (en) | Sequence image weak and small target detection method based on time domain filtering | |
| Dong et al. | Infrared target detection in backlighting maritime environment based on visual attention model | |
| Dehghani et al. | Small target detection and tracking based on the background elimination and Kalman filter | |
| CN104199009B (en) | A kind of radar image clutter suppression method based on time domain specification | |
| CN112669332B (en) | Method for judging sea-sky conditions and detecting infrared targets based on bidirectional local maxima and peak value local singularities | |
| CN107256559B (en) | Method for complex background suppression | |
| CN115631114B (en) | Dark and weak moving target indication enhanced on-track processing method based on time domain profile analysis | |
| Jiang et al. | An infrared small target detection algorithm based on peak aggregation and Gaussian discrimination | |
| Qin et al. | A background extraction and shadow removal algorithm based on clustering for ViBe | |
| Li et al. | Algorithm of Canny Operator Edge Pre-processing Based on Mathematical Morphology | |
| US11145040B2 (en) | Dynamic background clutter suppression | |
| Wang et al. | Image edge detection algorithm based onwavelet fractional differential theory | |
| Sun et al. | Infrared small targets detection based on MRF model | |
| Ning et al. | Ship detection of infrared image in complex scene based on bilateral filter enhancement | |
| Peng et al. | Infrared small target detection under complex background | |
| Zhang et al. | Infrared small target detection based on anisotropic contrast filter | |
| Wen et al. | A method for automatic infrared point target detection in a sea background based on morphology and wavelet transform | |
| Khan et al. | An image enhancement technique of x-ray carry-on luggage for detection of contraband/illicit Object (s) | |
| CN107833195A (en) | The in-orbit dark weak signal target optical imagery Enhancement Method in ocean | |
| Jie et al. | A bullet holes detection algorithm based on wavelet transform and image fusion | |
| CN115187918B (en) | Method and system for identifying moving object in monitoring video stream |
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 |