CN106875335A - It is a kind of improved based on Hilbert-Huang transform image background suppressing method - Google Patents
It is a kind of improved based on Hilbert-Huang transform image background suppressing method Download PDFInfo
- Publication number
- CN106875335A CN106875335A CN201710084962.3A CN201710084962A CN106875335A CN 106875335 A CN106875335 A CN 106875335A CN 201710084962 A CN201710084962 A CN 201710084962A CN 106875335 A CN106875335 A CN 106875335A
- Authority
- CN
- China
- Prior art keywords
- image
- background
- pixel
- hilbert
- huang transform
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 20
- 230000009466 transformation Effects 0.000 claims abstract description 8
- 238000001514 detection method Methods 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 description 6
- 230000001629 suppression Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/20—Linear translation of whole images or parts thereof, e.g. panning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
Improved based on Hilbert-Huang transform image background suppressing method the invention discloses one kind, its step is:(1)Gradation of image linear transformation.Each pixel gray value linear transformation in image F (x, y), to the space of [0,255], is saved as into image G (x, y);(2)Seek extreme point.The maximum point and minimum point in image G (x, y) are found out, corresponding pixel position and gray value are stored in maximum point set and minimum point set respectively;(3)Residual image is calculated with Hilbert-Huang transform.Wherein, (x, y) represents the position coordinates of pixel, and 0≤x<M 1,0≤y<N 1, F (x, y) are the gray value of (x, y).It is an advantage of the invention that:Solve the problems, such as existing method be easily caused residual image in Dim targets detection in Weak target signal it is fainter, as far as possible retain target information simultaneously can effectively suppress noise, improve the signal to noise ratio of image.
Description
Technical field
The invention belongs to areas of information technology, the Dim targets detection that can be used in single-frame images is specially a kind of to improve
Based on Hilbert-Huang transform image background suppressing method.
Background technology
Background suppression technology is most efficient method in the target detection based on single-frame images, and its basic thought is:One
Pair has powerful connections in the image of interference, and the correlation using the gray value of gray value and the surrounding pixel point of pixel in background compares
Greatly, the general presence in the form of isolated point of target, its correlation very little with ambient background pixel.Using this property,
We can predict the gray scale of the pixel with the average value of a range of pixel around each pixel in image
Value, this process we be referred to as background forecast.Due to the gray value of the actual grey value of pixel and surrounding pixel point in background
Correlation than larger, the predicted value of the gray value of pixel and the pixel can be very close in background.And target pixel points
The gray value correlation very little of the pixel of gray value and ambient background, target pixel points actual grey value and the pixel it is pre-
Measured value differs greatly.So subtracting its corresponding background forecast image with original image, suppression background, prominent target can be reached
Effect.
The key of background suppression technology be exactly make the predicted value maximum possible of the gray value of pixel in background close to its
Actual grey value, and make the predicted value maximum possible of the gray value of the pixel of target much smaller than its actual gray value so that
So that background maximum can during original image F (x, y) subtracts residual image E (x, y) after corresponding background forecast image B (x, y)
Can be eliminated, target maximum may be highlighted.
The basic mathematic model of background forecast technology can be expressed as:
B (x, y) is the background forecast result of present image in formula, and W (i, j) is the k × k centered on pixel (i, j)
Background forecast template, k is positive integer, SkIt is the pixel point set in background forecast template, general satisfaction ∑ W (i, j)=1.
Residual image E (x, y) can be expressed as:E (x, y)=F (x, y)-B (x, y) (2)
When background forecast is carried out to the image containing Weak target, if the gray scale of impact point is differed not with background gray scale
When big, background forecast directly is carried out using simple template, the signal for being easily caused Weak target in residual image will be more micro-
It is weak, cause the detection of Weak target more difficult.
The content of the invention
In the method for above-mentioned background technology, although the background of image can be predicted, but when the gray scale of impact point
When being more or less the same with background gray scale, the signal for being easily caused Weak target in residual image is fainter, causes the inspection of Weak target
Survey more difficult;The purpose of the present invention is directed to above mentioned problem and proposes solution.
Based on described above, it is proposed that a kind of new departure is improved to above-mentioned mechanism;It is a kind of improved based on Xi Er
Bert Huang image background suppressing method, infrared digital image F (x, y) for M × N pixels (M, N are positive integer) changes
The process based on Hilbert-Huang transform image background suppressing method entered is largely divided into three steps, wherein, (x, y) represents pixel
Position coordinates, and 0≤x<M-1,0≤y<N-1, F (x, y) are the gray value of (x, y);It is characterized in that method and step is as follows:
(1) gradation of image linear transformation.By sky of each pixel gray value linear transformation to [0,255] in image F (x, y)
Between, save as image G (x, y);
(2) extreme point is sought.Find out the maximum point and minimum point in image G (x, y), by corresponding pixel position and
Gray value is stored in maximum point set and minimum point set respectively;
(3) residual image is calculated with Hilbert-Huang transform.
Its step is as follows in step (3) of the present invention:
1. extreme value dot image is reconstructed;Two dimension is carried out to maximum point set and minimum point set with RBF to insert
Value, reconstructs maximum point image and minimum point image;
2. background forecast image is calculated;Background forecast image B (x, y) of image G (x, y) is maximum point image and minimum
It is worth the half of dot image sum;
3. residual image is calculated;Residual image E (x, y) of image G (x, y) subtracts background forecast figure equal to image G (x, y)
As B (x, y), that is, obtain the residual image after background suppresses.
It is an advantage of the invention that:Solve Weak target during existing method is easily caused residual image in Dim targets detection
The fainter problem of signal, propose it is a kind of improved based on Hilbert-Huang transform image background suppressing method, as far as possible
Retaining target information can effectively suppress noise simultaneously, improve the signal to noise ratio of image.
The possible range of application of this patent:The detection of Weak target in digital picture.
Specific embodiment
Below application process of this patent in digital picture background suppresses is illustrated with example.
For image F (x, y) of M × N pixels (M, N are positive integer), wherein, (x, y) represents the position coordinates of pixel,
And 0≤x<M-1,0≤y<N-1, F (x, y) are the gray value of pixel (x, y).Comprise the following steps that:
(1) grey linear transformation.
By each pixel gray value linear transformation in image F (x, y) to the space of [0,255], image G (x, y) is saved as, counted
Calculating formula is
Wherein, FmaxIt is the maximum of gray value in image F (x, y), corresponding pixel is designated as (xmax,ymax);FminIt is
The minimum value of gray value in image F (x, y), corresponding pixel is designated as (xmin,ymin)。
(2) extreme point is sought.Find out the maximum point and minimum point in image G (x, y), by corresponding pixel position and
Gray value is stored in maximum point set and minimum point set respectively.Its step is as follows:
1. with 8 neighborhood windows from the pixel (x in image G (x, y)max,ymax) centered on point start, traversing graph as G (x,
y);When the gray value of any one pixel during the gray value of central point is not less than 8 neighborhoods, the central point is local pole
Big value point;Output image BW1 is bianry image, and local maximum is set to 1, and other pixel values are set as 0.
2. with 8 neighborhood windows from the pixel (x in image G (x, y)min,ymin) centered on point start, traversing graph as G (x,
y);When the gray value of central point is not more than the gray value of any one pixel in 8 neighborhoods, the central point is local pole
Small value point;Output image BW2 is bianry image, and local minimum is set to 1, and other pixel values are set as 0.
3. for the pixel in the 1 corresponding image G (x, y) of pixel constitutes maximum point set in BW1;It is 1 in BW2
The corresponding image G (x, y) of pixel in pixel constitute minimum point set.
(3) residual image is calculated with Hilbert-Huang transform.Its step is as follows:
1. extreme value dot image is reconstructed.Two dimension is carried out to maximum point set and minimum point set with MQ RBFs to insert
Value, reconstruct maximum point image Max (x, y) and minimum point image Min (x, y).MQ RBFs are that Hardy is proposed
Application RBF widely is planted, approximation accuracy is outstanding.
2. prognostic chart picture is calculated.Background forecast image B (x, y) of image G (x, y) be maximum point image Max (x, y) with
The half of minimum point image Min (x, y) sum, i.e.,
3. residual image is calculated.Residual image E (x, y) of image G (x, y) subtracts background forecast figure equal to image G (x, y)
As B (x, y), that is, obtain the residual image after background suppresses
E (x, y)=G (x, y)-B (x, y) (7).
Claims (2)
1. a kind of improved based on Hilbert-Huang transform image background suppressing method, for M × N pixels(M, N are positive integer)
The improved process based on Hilbert-Huang transform image background suppressing method of infrared digital image F (x, y) be largely divided into
Three steps, wherein, (x, y) represents the position coordinates of pixel, and 0≤x<M-1,0≤y<N-1, F (x, y) are the gray scale of (x, y)
Value;It is characterized in that method and step is as follows:
(1)Gradation of image linear transformation, by the space of each pixel gray value linear transformation in image F (x, y) to [0,255],
Save as image G (x, y);
(2)Extreme point is sought, maximum point and minimum point in image G (x, y) is found out, by corresponding pixel position and gray scale
Value is stored in maximum point set and minimum point set respectively;
(3)Residual image is calculated with Hilbert-Huang transform.
2. one kind according to claim 1 is improved based on Hilbert-Huang transform image background suppressing method, its feature
It is:The step(3)In its step it is as follows:
1. extreme value dot image is reconstructed;Two-dimensional interpolation is carried out to maximum point set and minimum point set with RBF, weight
Structure maximum point image and minimum point image;
2. background forecast image is calculated;Background forecast image B (x, y) of image G (x, y) is maximum point image and minimum point
The half of image sum;
3. residual image is calculated;Residual image E (x, y) of image G (x, y) subtracts background forecast image B equal to image G (x, y)
(x, y), that is, obtain the residual image after background suppresses.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710084962.3A CN106875335A (en) | 2017-02-16 | 2017-02-16 | It is a kind of improved based on Hilbert-Huang transform image background suppressing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710084962.3A CN106875335A (en) | 2017-02-16 | 2017-02-16 | It is a kind of improved based on Hilbert-Huang transform image background suppressing method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106875335A true CN106875335A (en) | 2017-06-20 |
Family
ID=59166090
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710084962.3A Pending CN106875335A (en) | 2017-02-16 | 2017-02-16 | It is a kind of improved based on Hilbert-Huang transform image background suppressing method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106875335A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115859405A (en) * | 2023-03-02 | 2023-03-28 | 青岛昊宇重工有限公司 | Design data enhancement method for self-supporting steel chimney |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101359402A (en) * | 2008-09-11 | 2009-02-04 | 北京航空航天大学 | Automatic selection method of infrared background forecasting operator under complex background |
US20110052088A1 (en) * | 2009-08-31 | 2011-03-03 | Yuan Xiaoru | High dynamic range image mapping with empirical mode decomposition |
CN102184529A (en) * | 2011-05-12 | 2011-09-14 | 西安电子科技大学 | Empirical-mode-decomposition-based edge detecting method |
CN103500440A (en) * | 2013-09-27 | 2014-01-08 | 中国人民解放军空军工程大学 | Method for eliminating cloud and haze of atmospheric degraded image |
CN104103080A (en) * | 2014-07-02 | 2014-10-15 | 华中科技大学 | Method of small dim target detection under complicated background |
CN106096571A (en) * | 2016-06-22 | 2016-11-09 | 北京化工大学 | A kind of based on EMD feature extraction with the cell sorting method of rarefaction representation |
-
2017
- 2017-02-16 CN CN201710084962.3A patent/CN106875335A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101359402A (en) * | 2008-09-11 | 2009-02-04 | 北京航空航天大学 | Automatic selection method of infrared background forecasting operator under complex background |
US20110052088A1 (en) * | 2009-08-31 | 2011-03-03 | Yuan Xiaoru | High dynamic range image mapping with empirical mode decomposition |
CN102184529A (en) * | 2011-05-12 | 2011-09-14 | 西安电子科技大学 | Empirical-mode-decomposition-based edge detecting method |
CN103500440A (en) * | 2013-09-27 | 2014-01-08 | 中国人民解放军空军工程大学 | Method for eliminating cloud and haze of atmospheric degraded image |
CN104103080A (en) * | 2014-07-02 | 2014-10-15 | 华中科技大学 | Method of small dim target detection under complicated background |
CN106096571A (en) * | 2016-06-22 | 2016-11-09 | 北京化工大学 | A kind of based on EMD feature extraction with the cell sorting method of rarefaction representation |
Non-Patent Citations (2)
Title |
---|
ZHONGMIN LI ET AL.: "Research on Three Typical Methods of Infrared Weak Small Targets Background Suppression under Cloud Background", 《INTERNATIONAL CONFERENCE ON MATERIALS ENGINEERING AND INFORMATION TECHNOLOGY APPLICATIONS (MEITA 2015)》 * |
张丽珂: "基于相似性滤波的红外小目标检测", 《微计算机应用》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115859405A (en) * | 2023-03-02 | 2023-03-28 | 青岛昊宇重工有限公司 | Design data enhancement method for self-supporting steel chimney |
CN115859405B (en) * | 2023-03-02 | 2023-05-12 | 青岛昊宇重工有限公司 | Self-supporting steel chimney design data enhancement method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8639056B2 (en) | Contrast enhancement | |
Li et al. | Single image haze removal using content‐adaptive dark channel and post enhancement | |
CN101640801B (en) | Image processing apparatus, image processing method | |
CN108564597B (en) | Video foreground object extraction method fusing Gaussian mixture model and H-S optical flow method | |
US20150339806A1 (en) | Image denoising method and image denoising apparatus | |
CN105654438A (en) | Gray scale image fitting enhancement method based on local histogram equalization | |
CN106169173B (en) | Image interpolation method | |
CN110796622B (en) | Image bit enhancement method based on multi-layer characteristics of series neural network | |
US7903900B2 (en) | Low complexity color de-noising filter | |
CN109377464B (en) | Double-platform histogram equalization method for infrared image and application system thereof | |
CN111242860B (en) | Super night scene image generation method and device, electronic equipment and storage medium | |
CN112529854A (en) | Noise estimation method, device, storage medium and equipment | |
JP5237814B2 (en) | Multi-stage link method for adaptive motion vector sampling in video compression | |
CN108270945B (en) | Motion compensation denoising method and device | |
Tan et al. | A real-time video denoising algorithm with FPGA implementation for Poisson–Gaussian noise | |
Zuo et al. | Video Denoising Based on a Spatiotemporal Kalman‐Bilateral Mixture Model | |
CN115564683A (en) | Ship detection-oriented panchromatic remote sensing image self-adaptive enhancement method | |
JP2005524353A (en) | How to process digital images for small rate applications | |
CN106875335A (en) | It is a kind of improved based on Hilbert-Huang transform image background suppressing method | |
JP2019207603A (en) | Tone mapping processing method by maximizing edge intensity, and hdr video conversion device | |
Cao et al. | Single image motion deblurring with reduced ringing effects using variational Bayesian estimation | |
CN116012265B (en) | Infrared video denoising method and device based on time-space domain adaptive filtering | |
CN117078574A (en) | Image rain removing method and device | |
CN108492264B (en) | Single-frame image fast super-resolution method based on sigmoid transformation | |
CN114663307B (en) | Integrated image denoising system based on uncertainty network |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170620 |