CN109919865A - A kind of image multilayer detail enhancing method - Google Patents
A kind of image multilayer detail enhancing method Download PDFInfo
- Publication number
- CN109919865A CN109919865A CN201910127062.1A CN201910127062A CN109919865A CN 109919865 A CN109919865 A CN 109919865A CN 201910127062 A CN201910127062 A CN 201910127062A CN 109919865 A CN109919865 A CN 109919865A
- Authority
- CN
- China
- Prior art keywords
- window
- detail
- assistant
- image
- details
- 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.)
- Granted
Links
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a kind of image multilayer detail enhancing methods, comprising the following steps: S1: selected window is as detail window in original image；S2: choosing multiple assistant windows in detail window, and the size of the assistant window is respectively less than detail window；S3: the smooth enhancing window of generation is smoothed to detail window；Multiple assistant windows are smoothed and generate multiple smooth assistant windows；S4: detaching smooth enhancing window respectively from detail window and multiple smooth assistant windows form multiple details assistant windows；S5: the enhanced window of details will be used as after smoothly enhancing window and multiple details assistant window weighted superpositions.A kind of image multilayer detail enhancing method of the present invention accounts for low-frequency image details, and account for the image detail of high frequency, while can also eliminate partial noise, so that image detail enhancing result is more accurate by the way that above-mentioned steps are arranged.
Description
Technical field
The present invention relates to computer image processing technologies, and in particular to a kind of image multilayer detail enhancing method.
Background technique
Image enhancement refers to by the certain information specifically needed in prominent image, while weakening or removing certain do not need
Information processing method.Algorithm for image enhancement is segmented into two major classes substantially: spatial processing method and transform-domain processing method.Histogram
Traditional air space algorithm such as figure equilibrium, space filtering, image edge acuity mostly also promotes image while enhancing contrast and makes an uproar
Sound.Small echo Enhancement Method takes into account airspace and the frequency domain characteristic of picture signal, and multidirectional and multi-resolution characteristics are to solve sky
Domain contradictory problems opened up a new way, but this classical way does not fully take into account the nonlinear characteristic of vision, and image increases
Strong purpose is the visual effect for improving image to improve accuracy of identification.
Existing image detail enhancement method can not generally take the image detail of low frequency and high frequency into account simultaneously, cause image
Details enhancing result is not accurate enough, and noise is higher.
Summary of the invention
The technical problem to be solved by the present invention is to existing image detail enhancement methods, can not generally take low frequency into account simultaneously
With the image detail of high frequency, cause image detail enhancing result not accurate enough, noise is higher, and it is an object of the present invention to provide a kind of image is more
Layer detail enhancing method, solves the above problems.
The present invention is achieved through the following technical solutions:
A kind of image multilayer detail enhancing method, comprising the following steps: S1: selected window is as details in original image
Window；S2: choosing multiple assistant windows in detail window, and the size of the assistant window is respectively less than detail window；S3: to thin
Section window is smoothed the smooth enhancing window of generation；Multiple assistant windows are smoothed and generate multiple smooth auxiliary
Window；S4: detaching smooth enhancing window respectively from detail window and multiple smooth assistant windows form multiple details auxiliary windows
Mouthful；S5: the enhanced window of details will be used as after smoothly enhancing window and multiple details assistant window weighted superpositions.
Existing image detail enhancement method can not generally take the image detail of low frequency and high frequency into account simultaneously, cause image
Details enhancing result is not accurate enough, and noise is higher.The present invention is in application, design core of the invention is that original image is expressed as base
This component and details coefficients enhance details coefficients and obtain enhanced image, on this basis so the pass that the present invention designs
Key is how to obtain details coefficients.
The first selected window in original image of the present invention is exactly to be ready for as detail window, detail window said herein
Details enhancing window, detail window carry out details enhancing after splicing can be obtained by original image enhancing details after as a result,
Then multiple assistant windows are chosen in detail window, the size of the assistant window is respectively less than detail window, multiple auxiliary windows
The effect of mouth is exactly to extract details, and the size of in general all assistant windows is all different, and can thus be covered more
It is tiny to can be very good extraction high frequency when this assistant window is lesser for the assistant window of kind size, inventor's discovery
Image detail, such as image texture details, but be detrimental to extract the lower big image detail of frequency, such as image border；And
When this assistant window is larger, then the lower big image detail of frequency can be extracted very well, be but unfavorable for extracting high frequency
Tiny image detail；So may be implemented to extract different frequency in image by different assistant windows using multiple assistant windows
Characteristics of image, achieve the purpose that Image Multiscale enhance.
In order to carry out detail extraction to detail window, detail window is smoothed the smooth enhancing window of generation, and
Multiple assistant windows are smoothed and generate multiple smooth assistant windows, functioning as being smoothed here will
Details in detail window and multiple assistant windows has carried out filtering erasing, to generate image as background image；At this
In the process, some noises unconspicuous in the picture can by being retained in background image after filtering parameter is adjusted,
Would not occur the enhanced situation of noise in this way in subsequent arithmetic to occur.
In order to realize that detail extraction, the present invention detach smooth enhancing window and multiple smooth auxiliary respectively from detail window
Window forms multiple details assistant windows；It is mentioned here to detach, refer to and subtraction is done to two images, it will be suitable in detail window
After the part of assistant window background detaches away, remaining is exactly the detail section that can be enhanced.Details assists window
The quantity that the quantity of mouth is equivalent to smooth assistant window adds one.
Then the enhanced window of details will be used as after smoothly enhancing window and multiple details assistant window weighted superpositions.This
Kind of way, which is equivalent in smooth background based on by detail window, enhances details, and when enhancing can be according to high frequency
The difference of details and low frequency details demand changes the weighted value of different details assistant windows, so that the present invention is more applicable in
Property.The present invention accounts for low-frequency image details, and account for the image detail of high frequency by setting above-mentioned steps, simultaneously also
Partial noise can be eliminated, so that image detail enhancing result is more accurate.
Further, the central point of the multiple assistant window is overlapped with the central point of detail window.
The present invention is in application, in order to enable details enhancing more synchronizes, by the central point and details of multiple assistant windows
The central point of window is set as being overlapped.
Further, further comprising the steps of: S6: detail window is slided in original image, and executes S1 until former
All points are all used as detail window central point to use in beginning image；S7: all enhanced window center points of details are spelled
It connects to form the enhanced image of details.
The present invention first carries out a details in application, in order to realize that the details to entire original image carries out multilayer enhancing
The multilayer details of window enhances, and detail window is then slided a pixel, executes the enhancing of multilayer details again, and so circulation is straight
It is completed the enhancing of multilayer details to entire original image, forms complete details enhancing after then splicing to these images
Original image afterwards.
Further, the detail window is the square window of m × m, and m is odd number；The assistant window is n × n
Square window, and n is odd number less than m.
The present invention is in application, set m for detail window to preferably be positioned to detail window and assistant window
The square window of × m, and set assistant window to the square window of n × n, since n and m is odd number, so one
The central point of square window is clearly determining.
Further, the size of the multiple assistant window is sequentially increased.
The present invention is in application, since the size of multiple assistant windows is sequentially increased, it is possible to realize multi-level details
Enhancing, when detail window is little a, it might even be possible to assistant window is all arranged using every rank, enhanced complete in from low to high
Whole details.
Further, detail window is smoothed described in step S3 and is carried out using the gray value to detail window
Filtering；It is described that multiple assistant windows are smoothed using being filtered to the gray value of assistant window.
The present invention is in application, be filtered to the gray value of detail window and be filtered to the gray value of assistant window can
To obtain background image well.
Further, the filtering uses gaussian filtering, bilateral filtering, guiding filtering and/or low-pass filtering.
Further, step S5 includes following sub-step: the weighted value of multiple details assistant windows is identical, and multiple details
The weighted value of assistant window is greater than the weighted value of smooth enhancing window.
The present invention is in application, in order to simplify the actual operation amount of multilayer details enhancing, by the power of multiple details assistant windows
Weight values are set as identical, in this way on the basis of guaranteeing the enhancing of multilayer details, can reduce by 40% or more operand.
The weighted value of the further smooth enhancing window takes 1, and the weighted value of the multiple details assistant window takes 2~
5。
Compared with prior art, the present invention having the following advantages and benefits:
A kind of image multilayer detail enhancing method of the present invention accounts for low-frequency image details by setting above-mentioned steps,
The image detail of high frequency is accounted for again, while can also eliminate partial noise, so that image detail enhancing result is more accurate.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below with reference to embodiment, the present invention is made
Further to be described in detail, exemplary embodiment of the invention and its explanation for explaining only the invention, are not intended as to this
The restriction of invention.
Embodiment 1
A kind of image multilayer detail enhancing method of the present invention, comprising the following steps: S1: selected window is made in original image
For detail window；S2: choosing multiple assistant windows in detail window, and the size of the assistant window is respectively less than detail window；
S3: the smooth enhancing window of generation is smoothed to detail window；It is multiple that generation is smoothed to multiple assistant windows
Smooth assistant window；S4: detaching smooth enhancing window respectively from detail window and multiple smooth assistant windows formed it is multiple thin
Save assistant window；S5: the enhanced window of details will be used as after smoothly enhancing window and multiple details assistant window weighted superpositions.
When the present embodiment is implemented, design core of the invention is that original image is expressed as fundametal component and details coefficients,
Enhance details coefficients on the basis of this and obtain enhanced image, so the key that the present invention designs is how to obtain details point
Amount.The first selected window in original image of the present invention is exactly to be ready for details as detail window, detail window said herein
The window of enhancing, detail window carry out details enhancing after splicing can be obtained by original image enhancing details after as a result, then
Multiple assistant windows are chosen in detail window, the size of the assistant window is respectively less than detail window, multiple assistant windows
Effect is exactly to extract details, and the size of in general all assistant windows is all different, and can thus cover a variety of rulers
Very little assistant window, inventor's discovery can be very good to extract the tiny image of high frequency when this assistant window is lesser
Details, such as image texture details, but be detrimental to extract the lower big image detail of frequency, such as image border；And work as this
When kind assistant window is larger, then the lower big image detail of frequency can be extracted very well, it is tiny to be but unfavorable for extraction high frequency
Image detail；So the figure for extracting different frequency in image by different assistant windows may be implemented using multiple assistant windows
As feature, achieve the purpose that Image Multiscale enhances.
In order to carry out detail extraction to detail window, detail window is smoothed the smooth enhancing window of generation, and
Multiple assistant windows are smoothed and generate multiple smooth assistant windows, functioning as being smoothed here will
Details in detail window and multiple assistant windows has carried out filtering erasing, to generate image as background image；At this
In the process, some noises unconspicuous in the picture can by being retained in background image after filtering parameter is adjusted,
Would not occur the enhanced situation of noise in this way in subsequent arithmetic to occur.
In order to realize that detail extraction, the present invention detach smooth enhancing window and multiple smooth auxiliary respectively from detail window
Window forms multiple details assistant windows；It is mentioned here to detach, refer to and subtraction is done to two images, it will be suitable in detail window
After the part of assistant window background detaches away, remaining is exactly the detail section that can be enhanced.It then will be smooth
The enhanced window of details is used as after enhancing window and multiple details assistant window weighted superpositions.This way is equivalent to thin
Details is enhanced in smooth background based on section window, when enhancing can be according to high frequency detail and low frequency details demand
Difference, change the weighted value of different details assistant windows, so that the present invention has more applicability.The present invention passes through in setting
Step is stated, that is, accounts for low-frequency image details, and accounts for the image detail of high frequency, while partial noise can also be eliminated, is made
It is more accurate to obtain image detail enhancing result.
Embodiment 2
The present embodiment on the basis of embodiment 1, the central point of the multiple assistant window and the central point of detail window
It is overlapped.
When the present embodiment is implemented, in order to enable details enhancing is more synchronous, by the central point of multiple assistant windows and carefully
The central point of section window is set as being overlapped.
Embodiment 3
The present embodiment is further comprising the steps of on the basis of embodiment 2: S6: detail window is sliding in original image
It is dynamic, and S1 is executed until point all in original image is all used as detail window central point to use；S7: all details are enhanced
Window afterwards is spliced to form the enhanced image of details.
When the present embodiment is implemented, in order to realize that the details to entire original image carries out multilayer enhancing, first carry out one thin
The multilayer details enhancing for saving window, then slides a pixel for detail window, executes the enhancing of multilayer details again, so recycles
Until entire original image is completed the enhancing of multilayer details, complete details is formed after then splicing to these images and is increased
Original image after strong.
Embodiment 4
For the present embodiment on the basis of embodiment 2, the detail window is the square window of m × m, and m is odd number；Institute
The square window that assistant window is n × n is stated, and n is the odd number less than m.
When the present embodiment is implemented, in order to preferably be positioned to detail window and assistant window, detail window is arranged
For the square window of m × m, and assistant window is set to the square window of n × n, since n and m is odd number, so one
The central point of a square window is clearly determining.
Embodiment 5
On the basis of embodiment 1, the size of the multiple assistant window is sequentially increased the present embodiment.
When the present embodiment is implemented, since the size of multiple assistant windows is sequentially increased, it is possible to realize multi-level thin
Section enhancing, when detail window is little a, it might even be possible to assistant window is all arranged using every rank, enhanced in from low to high
Complete details.
Embodiment 6
The present embodiment on the basis of embodiment 1, is smoothed using to thin detail window described in step S3
The gray value of section window is filtered；It is described to multiple assistant windows be smoothed using to the gray value of assistant window into
Row filtering.
When the present embodiment is implemented, the gray value of detail window is filtered and the gray value of assistant window is filtered
It can be very good to obtain background image.
Embodiment 7
On the basis of embodiment 1, step S5 includes following sub-step to the present embodiment: the weight of multiple details assistant windows
It is worth identical, and the weighted value of multiple details assistant windows is greater than the weighted value of smooth enhancing window.
When the present embodiment is implemented, in order to simplify the actual operation amount of multilayer details enhancing, by multiple details assistant windows
Weighted value is set as identical, in this way on the basis of guaranteeing the enhancing of multilayer details, can reduce by 40% or more operand.
Embodiment 8
For the present embodiment on the basis of Examples 1 to 7, Schilling original image is the image of a q × p, and detail window
The window that image_i is one 11 × 11, defines S11 (x, y), x=0,1,3,4...q here；Y=0,1,2,3...p, x, y
It is integer, x indicates that row coordinate, y indicate column coordinate, and S11 (x, y) indicates the field 11*11 centered on (x, y), so S11
(x, y) illustrates detail window.
In 11*11 detail window, segment out 3*3,5*5,7*7, tetra- assistant windows of 9*9, this four assistant windows and
Detail window is all the same central point, and each window can indicate the details of different frequency range difference characteristic；
Then four assistant windows and detail window are carried out with the gaussian filtering of image grayscale:
Do_3 is smooth assistant window after the filtering of 3*3 assistant window；
Do_5 is smooth assistant window after the filtering of 5*5 assistant window；
Do_7 is smooth assistant window after the filtering of 7*7 assistant window；
Do_9 is smooth assistant window after the filtering of 9*9 assistant window；
Do_11 is smoothly to enhance window after detail window filters；
Then detail extraction is carried out:
Detail3=image_i-Do_3；
Detail5=image_i-Do_5；
Detail7=image_i-Do_7；
Detail9=image_i-Do_9；
Detail11=image_i-Do_11；
Here Detail3, Detail5, Detail7, Detail9 and Detail11 is details assistant window；
Then details enhancing is carried out:
Dout_enhance=Do_11+gain* (Detail11+Detail9+Detail7+Detail5+Detail3)；
Dout_enhance is the enhanced window of details in above formula, and gain is gain coefficient, also corresponds to details auxiliary
The weighted value of smooth enhancing window is taken 1, gain to take 2 by the weighted value of window here.
Above-described specific embodiment has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that being not intended to limit the present invention the foregoing is merely a specific embodiment of the invention
Protection scope, all within the spirits and principles of the present invention, any modification, equivalent substitution, improvement and etc. done should all include
Within protection scope of the present invention.
Claims (9)
1. a kind of image multilayer detail enhancing method, which comprises the following steps:
S1: selected window is as detail window in original image；
S2: choosing multiple assistant windows in detail window, and the size of the assistant window is respectively less than detail window；
S3: the smooth enhancing window of generation is smoothed to detail window；Generation is smoothed to multiple assistant windows
Multiple smooth assistant windows；
S4: detaching smooth enhancing window respectively from detail window and multiple smooth assistant windows form multiple details auxiliary windows
Mouthful；
S5: the enhanced window of details will be used as after smoothly enhancing window and multiple details assistant window weighted superpositions.
2. a kind of image multilayer detail enhancing method according to claim 1, which is characterized in that the multiple assistant window
Central point be overlapped with the central point of detail window.
3. a kind of image multilayer detail enhancing method according to claim 2, which is characterized in that further comprising the steps of:
S6: detail window is slided in original image, and executes S1 until point all in original image is all used as details window
Mouth central point uses；
S7: all enhanced window center points of details are spliced to form the enhanced image of details.
4. a kind of image multilayer detail enhancing method according to claim 2, which is characterized in that the detail window is m
The square window of × m, and m is odd number；The assistant window is the square window of n × n, and n is the odd number less than m.
5. a kind of image multilayer detail enhancing method according to claim 1, which is characterized in that the multiple assistant window
Size be sequentially increased.
6. a kind of image multilayer detail enhancing method according to claim 1, which is characterized in that thin described in step S3
Section window is smoothed to use and be filtered to the gray value of detail window；It is described that multiple assistant windows are smoothly located
Reason is used and is filtered to the gray value of assistant window.
7. a kind of image multilayer detail enhancing method according to claim 6, which is characterized in that the filtering uses Gauss
Filtering, bilateral filtering, guiding filtering and/or low-pass filtering.
8. a kind of image multilayer detail enhancing method according to claim 1, which is characterized in that step S5 includes following son
Step:
The weighted value of multiple details assistant windows is identical, and the weighted value of multiple details assistant windows is greater than smooth enhancing window
Weighted value.
9. a kind of image multilayer detail enhancing method according to claim 8, which is characterized in that the smooth enhancing window
Weighted value take 1, the weighted value of the multiple details assistant window takes 2~5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910127062.1A CN109919865B (en) | 2019-02-20 | 2019-02-20 | Image multi-layer detail enhancement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910127062.1A CN109919865B (en) | 2019-02-20 | 2019-02-20 | Image multi-layer detail enhancement method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109919865A true CN109919865A (en) | 2019-06-21 |
| CN109919865B CN109919865B (en) | 2022-10-11 |
Family
ID=66961838
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910127062.1A Active CN109919865B (en) | 2019-02-20 | 2019-02-20 | Image multi-layer detail enhancement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109919865B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070223834A1 (en) * | 2006-03-23 | 2007-09-27 | Samsung Electronics Co., Ltd. | Method for small detail restoration in digital images |
| CN105957030A (en) * | 2016-04-26 | 2016-09-21 | 成都市晶林科技有限公司 | Infrared thermal imaging system image detail enhancing and noise inhibiting method |
| CN107274365A (en) * | 2017-06-15 | 2017-10-20 | 中国矿业大学(北京) | A kind of mine image intensification method based on unsharp masking and NSCT algorithms |
-
2019
- 2019-02-20 CN CN201910127062.1A patent/CN109919865B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070223834A1 (en) * | 2006-03-23 | 2007-09-27 | Samsung Electronics Co., Ltd. | Method for small detail restoration in digital images |
| CN105957030A (en) * | 2016-04-26 | 2016-09-21 | 成都市晶林科技有限公司 | Infrared thermal imaging system image detail enhancing and noise inhibiting method |
| CN107274365A (en) * | 2017-06-15 | 2017-10-20 | 中国矿业大学(北京) | A kind of mine image intensification method based on unsharp masking and NSCT algorithms |
Non-Patent Citations (2)
| Title |
|---|
| 张秋菊等: "基于自适应平滑滤波的图像增强", 《现代电子技术》 * |
| 赵秋宇等: "可增强图像细节的改进的直方图均衡化算法", 《信阳师范学院学报(自然科学版)》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109919865B (en) | 2022-10-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102800063B (en) | Image enhancement and abstraction method based on anisotropic filtering | |
| CN105654436A (en) | Backlight image enhancement and denoising method based on foreground-background separation | |
| CN108182671B (en) | Single image defogging method based on sky area identification | |
| Fang et al. | Single image dehazing and denoising with variational method | |
| WO2022141145A1 (en) | Object-oriented high-resolution remote sensing image multi-scale segmentation method and system | |
| WO2022088976A1 (en) | Image processing method and device | |
| CN111640128A (en) | Cell image segmentation method based on U-Net network | |
| Wang et al. | Enhancement for dust-sand storm images | |
| CN106846271A (en) | A kind of method of reticulate pattern in removal identity card picture | |
| CN109101985A (en) | It is a kind of based on adaptive neighborhood test image mismatch point to elimination method | |
| CN107993198A (en) | Optimize the image defogging method and system of contrast enhancing | |
| CN110660048A (en) | Leather surface defect detection algorithm based on shape characteristics | |
| CN106408533A (en) | Card image extraction method and card image extraction system | |
| CN112132848B (en) | Preprocessing method based on image layer segmentation and extraction | |
| Shao et al. | Edge-preserving image decomposition via joint weighted least squares | |
| CN109919865A (en) | A kind of image multilayer detail enhancing method | |
| CN108346128A (en) | A kind of method and apparatus of U.S.'s face mill skin | |
| CN110175972B (en) | Infrared image enhancement method based on transmission map fusion | |
| Xiong et al. | An adaptive bilateral filtering algorithm and its application in edge detection | |
| CN107038708A (en) | Application of the image recognition algorithm in paper-cut effect | |
| Shibata et al. | Reflection removal using RGB-D images | |
| Xu et al. | Adaptive Bilateral Texture Filter for Image Smoothing | |
| CN107085832A (en) | A kind of Fast implementation of the non local average denoising of digital picture | |
| CN113763404A (en) | Foam image segmentation method based on optimization mark and edge constraint watershed algorithm | |
| Yang | Evaluation of edge detection algorithm of frontal image of facial contour in plastic surgery |
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 |