CN108133461A - A kind of real-time video low-light (level) Enhancement Method - Google Patents
A kind of real-time video low-light (level) Enhancement Method Download PDFInfo
- Publication number
- CN108133461A CN108133461A CN201711292570.2A CN201711292570A CN108133461A CN 108133461 A CN108133461 A CN 108133461A CN 201711292570 A CN201711292570 A CN 201711292570A CN 108133461 A CN108133461 A CN 108133461A
- Authority
- CN
- China
- Prior art keywords
- image
- level
- light
- dark
- low
- 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 18
- 239000003595 mist Substances 0.000 claims abstract description 20
- 238000001914 filtration Methods 0.000 claims abstract description 16
- 230000002708 enhancing effect Effects 0.000 claims abstract description 9
- 238000005457 optimization Methods 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000005286 illumination Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004297 night vision Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/10024—Color 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
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to computer video processing technology field, more particularly to a kind of real-time video low-light (level) Enhancement Method.By being inverted to obtain pseudo- mist figure to inputted video image.The initial value of image dark channel is obtained by the local minimum filtering operation for carrying out carrying out O (1) computation complexity to pseudo- mist figure again after RGB triple channel minimum value calculates.The optimization for carrying out dark numerical value by having the function of the iterative gradient two-sided filter of Steerable filter.After the dark threshold parameter that video image is calculated by way of statistics with histogram, threshold statistical is carried out to raw video image, obtains air color parameter.Defogging puppet mist figure is reversely solved by atmospheric propagation model.Finally defogging puppet mist figure is inverted to obtain low-light (level) enhancing image.Enhance picture using the video low-light (level) that this method obtains, effectively increase the identification of information, improve the Operational Effectiveness Rat of combatant.
Description
Technical field
The invention belongs to computer video processing technology field, more particularly to a kind of real-time video low-light (level) enhancing side
Method.
Background technology
Airborne Video System source is under the extreme conditions such as low-light (level) and high dynamic range illumination, and image quality greatly declines, extremely
Influence perception and judgement of the human eye to environment.Traditional video low-light (level) Enhancement Method is higher due to its computation complexity, it is difficult to
Meet the airborne requirement handled in real time.And general simple algorithm for image enhancement can not meet the performance of video low-light (level) enhancing
It is required that.Under many scenes, low-light (level) enhancing is not played the role of to output foggy image, instead so that video pictures matter
Amount declines.Therefore, it is necessary to a kind of requirements that should meet low-light (level) enhancing performance, ensure that its computation complexity is relatively low and real again
The video low-light (level) Enhancement Method of Shi Xingneng higher.
Invention content
The purpose of the present invention is：
In order to solve the problem of the real-time low-light (level) enhancing of video under the conditions of the insufficient lights such as night vision or backlight, it is proposed that
A kind of real-time video low-light (level) Enhancement Method.
The technical scheme is that：
It is low in real time to carry out video image using the calculation of fixed complexity for a kind of real-time video low-light (level) Enhancement Method
Illumination enhancing operation, specifically includes following steps：
The first step inverts input low-light (level) video image, obtains pseudo- mist figure；
Second step, RGB triple channel is carried out to pseudo- mist figure, and minimum value solves pixel-by-pixel, obtains minimum value image；
Third walks, and the quick Local Minimum value filtering of parallelization is carried out to minimum value image, obtains dark primary value icon；
4th step to dark primary value icon by row piecemeal, carries out the Gradient Iteration filtering of parallelization, obtains dark row
Direction optimizes image；
5th step carries out column direction piecemeal to line direction optimization image, carries out to parallelization the Gradient Iteration filter of column direction
Wave obtains final dark channel image；
6th step, carries out final dark channel image statistics with histogram, and 1000 most bright pixels of setting are corresponding
Brightness number is dark threshold value；
7th step divides the pseudo- mist figure of input according to dark threshold value, only counts corresponding dark numerical value and is more than
The average color of the point of dark threshold value is air color parameter；
8th step is reversely solved according to dark physical model, obtains defogging puppet mist figure；
9th step inverts defogging puppet mist figure, obtains the enhanced video image of low-light (level) as output.
Present invention has the advantage that：
The real-time for realizing video low-light (level) algorithm accelerates, can be in embedded multi-core system or GPU multiple nucleus systems
In computation capability is made full use of to reach higher calculating speed-up ratio.This method becomes due to considering the details in video image
Change, ensure that detailed information is undistorted using the mode that iterative gradient filters, avoid the artificial traces such as halation in traditional algorithm,
Effect is calculated with preferable.Enhance picture using the video low-light (level) that the algorithm obtains, effectively increase the identification of information
Degree improves the Operational Effectiveness Rat of combatant.
Description of the drawings
Fig. 1 is the principle of the present invention block diagram.
Specific embodiment
By taking certain type DSP embedded systems as an example, this method is performed using multinuclear programming mode, is as follows：
The first step according to multi-core parallel concurrent, carries out image inversion to input low-light (level) image line by line, pseudo- mist figure is obtained.
Inverting implementation method is：
I (x, y)=255-I0(x,y)；
Second step according to multi-core parallel concurrent, is swept pseudo- mist figure to piecemeal point by point.It retouches, RGB triple channel is obtained most
Small value is stored as minimum value image, and the calculation formula of the minimum value of RGB triple channel is：
Third walks, and the Local Minimum value filtering of minimum value image is carried out using Van Herk mini-value filterings algorithm, due to
The separable characteristic of mini-value filtering, the column direction filtering of filtering and then progress that a line direction is carried out to image are asked
Go out region minimum value to store as dark primary value icon；The calculation formula of mini-value filtering is：
The filtering mode uses the Van Herk mini-value filtering algorithms of O (1) computation complexity, and is handled using piecemeal
Mode carried out parallel acceleration；
4th step carries out Steerable filter to input dark primary value icon, using the pseudo- mist figure of input as guiding texture image,
Parallel iteration gradient filtering is carried out according to line direction；The Steerable filter processing method is：For each pixel, current
It is only related to the output y [i-1] of current output x [i] and a upper pixel to export y [i]；I.e.：
Y [i]=(1-a) x [i]+ay [i-1]
Wherein a is feedback constant；
5th step carries out a column direction Steerable filter, equally using pseudo- mist figure as leading again to the image of the 4th step output
To texture, output dark optimization image.
6th step, to dark optimization image progress statistics with histogram, the minimum value of 1000 numerical value is made before histogram luminance existence
For dark threshold parameter, the dark after optimization is denoted as D (i), and carrying out statistics with histogram to dark obtains histogram
For：
hist(i),0≤i≤255；
After finding the peak of its Luminance Distribution, counted from high to low according to brightness histogram, work as count value
During more than 1000, stop counting, it is dark threshold value to think present intensity numerical value at this time；I.e.：
7th step divides the pseudo- mist figure of input according to dark threshold value, only counts corresponding dark numerical value and is more than
The average color of the point of dark threshold value is air color parameter；
Color Statistical is carried out to original input video image using the threshold value T that step 6 obtains, air color is at this time：
It is to deserved air constant：
8th step is reversely solved using the dark channel image and color parameter of optimization, calculates defogging figure；
After dark parameter D, air color parameter Ar, Ag, Ab and air constant A is acquired, dark object is utilized
Reason model can obtain coloured image after corresponding defogging；
Wherein Ir, Ig, Ib have mist video image for input, and Jr, Jg, Jb enhance video image for defogging；
9th step, defogging figure carry out the image after image inversion as output；Export low-light (level) enhancing video image JO(x,
Y)=255-J (x, y).
Claims (5)
1. a kind of real-time video low-light (level) Enhancement Method carries out video image low photograph in real time using the calculation of fixed complexity
Degree enhancing operation, it is characterised in that：Include the following steps：
The first step inverts input low-light (level) video image, obtains pseudo- mist figure；
Second step, RGB triple channel is carried out to pseudo- mist figure, and minimum value solves pixel-by-pixel, obtains minimum value image；
Third walks, and the quick Local Minimum value filtering of parallelization is carried out to minimum value image, obtains dark primary value icon；
4th step to dark primary value icon by row piecemeal, carries out the Gradient Iteration filtering of parallelization, obtains dark line direction
Optimize image；
5th step carries out column direction piecemeal to line direction optimization image, carries out to parallelization the Gradient Iteration filtering of column direction, obtain
To final dark channel image；
6th step carries out final dark channel image statistics with histogram, the most bright corresponding brightness of 1000 pixels of setting
Numerical value is dark threshold value；
7th step divides the pseudo- mist figure of input according to dark threshold value, and only the corresponding dark numerical value of statistics, which is more than, helps secretly
The average color of the point of road threshold value is air color parameter；
8th step is reversely solved according to dark physical model, obtains defogging puppet mist figure；
9th step inverts defogging puppet mist figure, obtains the enhanced video image of low-light (level) as output.
2. a kind of real-time video low-light (level) Enhancement Method according to claim 1, it is characterised in that：The reversion is realized public
Formula is：
I (x, y)=255-I0(x,y)。
3. a kind of real-time video low-light (level) Enhancement Method according to claim 1, it is characterised in that：The RGB threeway
The calculation formula of the minimum value in road is：
4. a kind of real-time video low-light (level) Enhancement Method according to claim 1, it is characterised in that：The mini-value filtering
Calculation formula be：
5. a kind of real-time video low-light (level) Enhancement Method according to claim 1, it is characterised in that：At the Steerable filter
Reason method is：For each pixel, current output y [i] only with current output x [i] and a upper pixel
Output y [i-1] it is related；I.e.：
Y [i]=(1-a) x [i]+ay [i-1]
Wherein a is feedback constant.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711292570.2A CN108133461A (en) | 2017-12-07 | 2017-12-07 | A kind of real-time video low-light (level) Enhancement Method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711292570.2A CN108133461A (en) | 2017-12-07 | 2017-12-07 | A kind of real-time video low-light (level) Enhancement Method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108133461A true CN108133461A (en) | 2018-06-08 |
Family
ID=62389265
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711292570.2A Pending CN108133461A (en) | 2017-12-07 | 2017-12-07 | A kind of real-time video low-light (level) Enhancement Method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108133461A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109523474A (en) * | 2018-10-19 | 2019-03-26 | 福州大学 | A kind of enhancement method of low-illumination image based on greasy weather degradation model |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103177424A (en) * | 2012-12-07 | 2013-06-26 | 西安电子科技大学 | Low-luminance image reinforcing and denoising method |
| CN105488769A (en) * | 2015-12-08 | 2016-04-13 | 中国航空工业集团公司西安航空计算技术研究所 | Real time video defogging method |
| CN106530249A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Low-illumination color image enhancement method based on physical model |
-
2017
- 2017-12-07 CN CN201711292570.2A patent/CN108133461A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103177424A (en) * | 2012-12-07 | 2013-06-26 | 西安电子科技大学 | Low-luminance image reinforcing and denoising method |
| CN105488769A (en) * | 2015-12-08 | 2016-04-13 | 中国航空工业集团公司西安航空计算技术研究所 | Real time video defogging method |
| CN106530249A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Low-illumination color image enhancement method based on physical model |
Non-Patent Citations (3)
| Title |
|---|
| XUAN DONG ET AL.: ""FAST EFFICIENT ALGORITHM FOR ENHANCEMENT OF LOW LIGHTING VIDEO"", 《IEEE》 * |
| 王小元 等: ""基于物理模型的低照度图像增强算法"", 《计算机应用》 * |
| 白林亭 等: ""多核DSP图像去雾优化方法"", 《电光与控制》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109523474A (en) * | 2018-10-19 | 2019-03-26 | 福州大学 | A kind of enhancement method of low-illumination image based on greasy weather degradation model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103593830B (en) | A kind of low illumination level video image enhancement | |
| Xu et al. | Fast image dehazing using improved dark channel prior | |
| CN103606182B (en) | Method and device for image rendering | |
| CN105046658B (en) | A kind of low-light (level) image processing method and device | |
| CN105488769B (en) | A kind of real-time video defogging method | |
| CN102231206B (en) | Colorized night vision image brightness enhancement method applicable to automotive assisted driving system | |
| CN108765342A (en) | A kind of underwater image restoration method based on improvement dark | |
| CN103714520B (en) | Digital video image strengthening system and its method are realized based on FPGA | |
| CN110706172B (en) | Low-illumination color image enhancement method based on adaptive chaotic particle swarm optimization | |
| CN108876743A (en) | A kind of image rapid defogging method, system, terminal and storage medium | |
| CN106886985A (en) | A kind of self adaptation enhancement method of low-illumination image for reducing colour cast | |
| CN105118027B (en) | A kind of defogging method of image | |
| CN104318535B (en) | The method, device and mobile terminal of image defogging | |
| CN103440674B (en) | A kind of rapid generation of digital picture wax crayon specially good effect | |
| CN108154492B (en) | A kind of image based on non-local mean filtering goes haze method | |
| CN108416745A (en) | Image self-adaptive defogging enhancement method with color constancy | |
| CN105959510B (en) | A kind of video rapid defogging method | |
| CN111145105B (en) | Image rapid defogging method and device, terminal and storage medium | |
| CN109493299B (en) | Method for eliminating illumination influence of point light source | |
| CN104331867B (en) | The method, device and mobile terminal of image defogging | |
| CN106485674A (en) | A kind of low light image Enhancement Method based on integration technology | |
| CN104200427A (en) | Method for eliminating edge sawteeth of images | |
| CN109523474A (en) | A kind of enhancement method of low-illumination image based on greasy weather degradation model | |
| CN102446345A (en) | Method, device and system for carrying out color enhancement on whiteboard colored image | |
| CN104881848A (en) | Parallel optimization method of low-illumination image enhancement based on CUDA |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180608 |