CN104156921B - Self-adaptive low-illuminance or non-uniform-brightness image enhancement method - Google Patents
Self-adaptive low-illuminance or non-uniform-brightness image enhancement method Download PDFInfo
- Publication number
- CN104156921B CN104156921B CN201410389246.2A CN201410389246A CN104156921B CN 104156921 B CN104156921 B CN 104156921B CN 201410389246 A CN201410389246 A CN 201410389246A CN 104156921 B CN104156921 B CN 104156921B
- Authority
- CN
- China
- Prior art keywords
- brightness
- image
- pretreatment
- color space
- pixel
- 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
Abstract
The invention relates to a self-adaptive low-illuminance or non-uniform-brightness image enhancement method. The method comprises the following steps: 1), preprocessing is performed on a low-illuminance and non-uniform-brightness image, wherein the preprocessing includes brightness preprocessing on the low-illuminance and non-uniform-brightness image, and edge enhancement is performed on the image after brightness preprocessing, so that the preprocessed image is obtained; 2), region segmentation is performed according to the brightness of the preprocessed image, corresponding mapping functions are selected according to the different characteristics of all segmented regions and corresponding self-adaptive brightness enhancement is performed; 3), saturation enhancement processing is performed on the image subjected to self-adaptive brightness enhancement segment by segment through the change characteristics of initial saturation and brightness. According to the invention, the steps are adopted to process the image, therefore, the color saturation of the image is improved, the image is enabled to be bright in color and have a better visual effect. The self-adaptive image enhancement method can be widely popularized in the fields of biomedicine, real-time monitoring, satellite remote sensing and the like.
Description
Technical field
The present invention relates to a kind of image enchancing method, the self adaptation of specifically a kind of low-light (level) or brightness disproportionation image
Image enchancing method.
Background technology
Image procossing application in, when environment light source illumination is relatively low or uneven illumination is even, often result in image detail unclear,
Information is lost or picture quality degradation, now it is necessary to improve picture quality with image enhancement technique, in order to follow-up
Image procossing.Image enhancement technique refers to the application scenario for given image, purposefully emphasizes entirety or the office of image
Portion's characteristic, unsharp image originally is apparent from, or emphasizes that some features interested suppress uninterested feature,
Difference between different objects feature in expanded view picture, improves picture quality, abundant information amount, strengthens image interpretation and identification effect
Really, some special analysis and the needs processing are met.
Conventional images enhancement techniques are broadly divided into transform domain method and airspace enhancement method, transform domain method be frequency domain,
Wavelet field or other transform domain carry out image enhaucament, be mainly used in the relatively low image of signal to noise ratio, but its computation complexity is higher；
Airspace enhancement method is then directly to carry out enhancement process in spatial domain to image, and its computation complexity is relatively low.Spatial domain enhancing side
Method is mainly by histogram equalizing method, restriction Contrast-limited adaptive histogram equalization method (CLAHE, Contrast
Limited Adaptive Histogram Equalization), γ-bearing calibration, retina cerebral cortex theoretical method
And brightness mapping function method of adjustment (Retinex).In these methods, brightness mapping function method of adjustment adaptability is good, spirit
Activity is high, is therefore used widely.
Brightness maps the selection that it is critical only that mapping function of method of adjustment, and the mapping function commonly used at present has sinusoidal letter
Number, exponential function, logarithmic function and parabolic function.In order that the method adapts to different images, draw in mapping function
Enter auto-adaptive parameter, common auto-adaptive parameter has brightness of image, standard deviation and image entropy etc., now illustrated and adopt these
The defect of auto-adaptive parameter：
1) Tao L, Asari V K. is published in《Journal of Electronic Imaging》, 2005,14 (4):
" the Adaptive and integrated neighborhood-dependent approach of 043006-043006-14.
For nonlinear enhancement of color images (AINDANE, the non-thread based on overall pixel and its neighborhood
Property self-adapting enhancement method) " in employ exponential function and standard deviation criteria, under the conditions of improving relatively low or brightness disproportionation
The visual quality of image.Image is converted into gray level image by the method first, then carries out adaption brightness increasing to this gray level image
Strong and Adaptive contrast enhancement.Adaption brightness strengthen be with custom-designed function come Automatic adjusument the overall situation brightness,
Largely increased the brightness compared with dark pixel, have compressed the dynamic range of image.Adaptive contrast enhancement is according to certain
The relative amplitude that individual pixel is adjacent pixel the brightness of this pixel is adjusted, and this process also adaptively passes through figure
The global statistics parameter of picture is controlled, and after having carried out this two steps, carries out color rendition to the image obtaining, will obtain
Gray level image be then converted to coloured image on the basis of original image property, this image be enhanced image.
Defect using said method is as follows：1. this technology is applied to low-light (level) image, but is not suitable for brightness irregularities
Image；2. this technology is related to convolution algorithm when calculating overall brightness distributed intelligence, and many places apply to complex exponent computing,
Amount of calculation is larger；3. brightness of image can be strengthened, but effectively can not keep colour vividness, make image visual effect be subject to shadow
Ring.
2) Ghimire D, Lee J. is published in《IEEE Transactions on Consumer Electronics》
2011,57 (2):" the Nonlinear transfer function-based local approach for of 858-865.
Color image enhancement (a kind of image enchancing method based on nonlinear transfer function and image local feature) "
In employ exponential function and brightness of image parameter, for contrast increasing is carried out to low-light (level) image and brightness irregularities image
By force.The method, first in HSV color space, carries out piecemeal enhancing with nonlinear transfer function to image, and piecemeal size also may be used
To be accurate to each pixel；Then, with the feature of center pixel and its surrounding pixel, Adaptive contrast enhancement is carried out to image.
Defect using said method is as follows：1. this technology, using fixing luminance segmentation parameter, is therefore divided for brightness
The image adaptability that cloth is extremely uneven, brightness fluctuation is larger is poor；2. this technology is related to when calculating Luminance Distribution information roll up
Long-pending computing, amount of calculation is larger；3. brightness of image can be strengthened, but effectively can not keep colour vividness, make image visual effect
It is affected.
In sum, still suffer from adaptivity using the regulation image enchancing method of common auto-adaptive parameter to be insufficient to
By force, the problems such as operand is big, color is not bright-coloured.
Content of the invention
For the problems referred to above, it is an object of the invention to provide the adapting to image of a kind of low-light (level) or brightness disproportionation image increases
Strong method.
For achieving the above object, the present invention takes technical scheme below：A kind of low-light (level) or brightness disproportionation image adaptive
Answer image enchancing method, it comprises the following steps：
1) pretreatment is done to low-light (level) and brightness disproportionation image, pretreatment includes first low-light (level) and brightness disproportionation image being done
Brightness pretreatment, and to brightness, pretreated image carries out edge enhancing, obtains pretreated image；
2) brightness according to pretreated image carries out region segmentation, and different special according to each region split
Point, selects corresponding mapping function, carries out corresponding adaption brightness enhancing respectively, to realize the adjustment of image overall brightness；
3) using the feature of initial saturation and brightness flop, to segmented adaptive brightness, enhanced image carries out saturation
Degree enhancement process, obtains final image.
Described step 1) comprise the following steps：
1. low-light (level) or brightness disproportionation image are transformed into HSV color space from rgb color space, to obtain HSV color
Original image in space；
Choose low-light (level) or brightness disproportionation image in rgb color space, this picture size is M × N, wherein, M is this image
Height, N is the width of this image；Any point coordinate in this image is (x, y), and x ∈ [0, M-1], y ∈ [0, N-1]；
For the pixel at any point (x, y) place, define R (x, y), G (x, y) and B (x, y) is in rgb color space
The red component of this pixel, green component and blue component；By the pixel of any point in low-light (level) or brightness disproportionation image from rgb color
Space is transformed into HSV color space, to realize for low-light (level) in rgb color space or brightness disproportionation image being transformed into HSV color
In space, obtain the original image in HSV color space, its each component of this pixel in HSV color space is：V (x, y) is
Luminance component, S (x, y) are saturation component and H (x, y) is chrominance component；
2. calculate the average brightness value V of the original image in HSV color spacea：
3. the threshold value of the average brightness value of original image and brightness pretreatment is compared, averagely bright when original image
Angle value is not more than the threshold value of brightness pretreatment, then carry out brightness pretreatment, otherwise, then do not process；
Threshold value V according to brightness pretreatmentthAverage brightness value V with original imageaBetween magnitude relationship decide whether
Need to carry out brightness pretreatment to the original image in HSV color space, its principle is as follows：
If the average brightness value V of original imageaIt is not more than threshold value V of brightness pretreatmentth, then adopt γ-alignment technique pair
Original image in HSV color space carries out brightness pretreatment；
Located in advance using the brightness that γ-alignment technique carries out brightness pretreatment gained to the original image in HSV color space
Luminance component image V after reason1(x,y)：
Wherein, γ1It is the parameter of γ-alignment technique, γ1Between (0,1), VthDepending on being needed according to experiment；
Using γ-alignment technique, original image is carried out after brightness pretreatment, the pretreated image of brightness of gained is bright
Degree component V1(x, y) automatically becomes the average brightness value V of original imagea, and threshold value V with brightness pretreatment by itthCompare,
If threshold value V of no more than brightness pretreatmentth, then again brightness pretreatment is carried out using γ-alignment technique, then with brightness pretreatment
Threshold value VthCompare, iterative cycles, until the average brightness value V of original imageaThreshold value V more than brightness pretreatmentthTill,
Processed subsequently into next step；If the average brightness value V of original imageaThreshold value V more than brightness pretreatmentth, then to HSV
Original image in color space does not carry out brightness pretreatment；
4. to brightness, pretreated image carries out edge enhancing, obtains pretreated image Vp(x,y).
Described step 2) comprise the following steps：
1. the brightness according to pretreated image carries out region segmentation, so that the difference according to each region split
Feature carries out corresponding adaption brightness enhancing；
If pretreated image VpThe mean flow rate of (x, y) is Va-p, by pretreated image VpAll pictures of (x, y)
Element is divided into two classes VlowAnd Vhigh：VlowRepresent the mean flow rate of low brightness pixel, VhighRepresent the mean flow rate of high luminance pixel,
Respective expression formula is as follows：
Wherein,Sgn () is sign sign function；
By pretreated image VpThe pixel of (x, y) is divided into three types according to brightness difference：Brightness is [0, Vlow) area
Between pixel be pretreated image VpThe low-light (level) area pixel of (x, y)；Brightness is in [Vlow,Vhigh) interval pixel is pre-
Image V after processpThe normal region pixel of (x, y)；Brightness is in [Vhigh, 255] and interval pixel is pretreated image Vp
The exposure area pixel of (x, y)；
2. the different characteristics according to each region split, select corresponding mapping function, carry out respectively corresponding from
Adaptation brightness strengthens, to realize the adjustment of image overall brightness；
Pretreated image VpThe brightness histogram corresponding to three area pixel that the pixel of (x, y) is divided, selects
Following mapping function carries out brightness of image enhancing respectively, and its formula is as follows：
Wherein, VenhPixel brightness value at coordinate points (x, y) place, log after (x, y) expression brightness enhancing10() is with 10
The logarithm bottom of for：
Wherein, γ2It is the parameter of γ-alignment technique, between (0,1).
Described step 3) include herein below：
Through the relatively pretreated image of the enhanced image of segmentation brightness (x, y) place pixel brightness flop Δ V
(x, y) is
Δ V (x, y)=[Venh(x,y)-Vp(x,y)]/255
Average staturation value S of the original image in HSV color spaceaFor
To segmentation brightness, enhanced image carries out saturation enhancement process, and its process is as follows：
Wherein, SenhThe pixel intensity value at (x, y) place for the image after (x, y) expression saturation enhancement process；
Image after saturation enhancement process is converted back rgb color space from HSV color space, obtains final cromogram
Picture.
Due to taking above technical scheme, it has advantages below to the present invention：1st, the present invention does original image including bright
Degree pretreatment and the enhanced pretreatment in edge, and the brightness of pretreated image is carried out region segmentation, according to split
The different characteristics in each region, selects corresponding mapping function, and the parameter of segmentation is determined by image own characteristic with principle, because
This adaptivity is preferably, more applicable particularly with the image that Luminance Distribution is extremely uneven, brightness fluctuation is larger.2nd, institute of the present invention
Multiplication, addition and the relatively low exponential function of power mostly are using formula operation mode, without complex calculation, reduce amount of calculation,
Thus improve image processing efficiency, it is easy to view synthesis.3rd, the present invention carries out saturation increasing after brightness of image enhancing
By force, and in image saturation strengthens, brightness of image change is combined with original image saturation, gives mutual system accordingly
About, the weight coefficient mutually balancing, so that image saturation strengthens has more preferable effect, and improves image color saturation
Degree, makes image color bright-coloured, has more preferable visual effect.For the foregoing reasons the present invention can biomedicine, monitor in real time,
The fields such as satellite remote sensing are widely popularized.
Brief description
Fig. 1 is the schematic flow sheet of the present invention
Fig. 2 a is the original image that rectangular histogram is a peak value
Fig. 2 b is that an original peak value histogram image is used alone with the image after γ-alignment technique is processed
Fig. 2 c is the image after an original peak value histogram image is used with CLAHE process
Fig. 2 d is the image after an original peak value histogram image is used with AINDANE process
Fig. 2 e is using the figure based on nonlinear transfer function and image local feature to an original peak value histogram image
Image after the process of image intensifying method
Fig. 2 f is the image after an original peak value histogram image is adopted with present invention process
Fig. 3 a is the original image that rectangular histogram is two peak values
Fig. 3 b is that original two peak value histogram image are used alone with the image after γ-alignment technique is processed
Fig. 3 c is the image after original two peak value histogram image are used with CLAHE process
Fig. 3 d is the image after original two peak value histogram image are used with AINDANE process
Fig. 3 e is using the figure based on nonlinear transfer function and image local feature to original two peak value histogram image
Image after the process of image intensifying method
Fig. 3 f is the image after original two peak value histogram image are adopted with present invention process
Fig. 4 a is the original image that rectangular histogram is three peak values
Fig. 4 b is that original three peak value histogram image are used alone with the image after γ-alignment technique is processed
Fig. 4 c is the image after original three peak value histogram image are used with CLAHE process
Fig. 4 d is the image after original three peak value histogram image are used with AINDANE process
Fig. 4 e is using the figure based on nonlinear transfer function and image local feature to original three peak value histogram image
Image after the process of image intensifying method
Fig. 4 f is the image after original three peak value histogram image are adopted with present invention process
Specific embodiment
The present invention is further detailed explanation with reference to the accompanying drawings and detailed description.
The present invention be directed to low-light (level) or brightness disproportionation image are processed, wherein low-light (level) image refers to overall brightness value
Relatively low image, and brightness disproportionation image refers to the image that Luminance Distribution is uneven, fluctuation is larger.
As shown in figure 1, the method for adaptive image enhancement of image low-light (level) of the present invention or brightness disproportionation image, it include with
Lower step：
1) pretreatment is done to low-light (level) and brightness disproportionation image, pretreatment includes first low-light (level) and brightness disproportionation image being done
Brightness pretreatment, and to brightness, pretreated image carries out edge enhancing, obtains pretreated image, it includes following step
Suddenly：
1. low-light (level) or brightness disproportionation image are transformed into HSV color space from rgb color space, so that more intuitively
Process (due to present invention is generally directed in HSV color space the brightness of image and saturation processed)；
Choose low-light (level) or brightness disproportionation image in rgb color space, this picture size is M × N, wherein, M is this image
Height, N is the width of this image.Any point coordinate in this image is (being also pixel) (x, y), and x ∈ [0, M-
1], y ∈ [0, N-1].
For the pixel at any point (x, y) place, define R (x, y), G (x, y) and B (x, y) is in rgb color space
The red component of this pixel, green component and blue component.Will be any one in low-light (level) or brightness disproportionation image by formula (1)～(3)
The pixel of point is transformed into HSV color space from rgb color space, to realize low-light (level) in rgb color space or brightness disproportionation
Image is transformed in HSV color space, obtains the original image in HSV color space, and in HSV color space, this pixel is bright
Degree component V (x, y), saturation component S (x, y) and chrominance component H (x, y) are as follows：
V (x, y)=[R (x, y)+G (x, y)+B (x, y)]/3 (1)
S (x, y)=1-min [R (x, y), G (x, y), B (x, y)]/V (x, y) (2)
Wherein, R (x, y) ∈ [0,255], G (x, y) ∈ [0,255], B (x, y) ∈ [0,255], according to above-mentioned formula (1)
Draw V (x, y) ∈ [0,255], S (x, y) ∈ [0,1] drawn according to above-mentioned formula (2), according to above-mentioned formula (3) draw H (x,
Y) ∈ [0,2 π], min [] function is the minima calculating all elements in [], and arccos [] is inverse cosine function.
2. calculate the average brightness value V of the original image in HSV color spacea：
3. brightness pretreatment is carried out to the original image in HSV color space；
Threshold value V according to brightness pretreatmentthAverage brightness value V with original imageaBetween magnitude relationship decide whether
Need to carry out brightness pretreatment to the original image in HSV color space, its principle is as follows：
If the average brightness value V of original imageaIt is not more than threshold value V of brightness pretreatmentth, then adopt γ-alignment technique pair
Original image in HSV color space carries out brightness pretreatment, subsequently to carry out effect is significant when brightness strengthens.
Located in advance using the brightness that γ-alignment technique carries out brightness pretreatment gained to the original image in HSV color space
Luminance component image V after reason1(x,y)：
Wherein, γ1It is the parameter of γ-alignment technique, γ1Between (0,1), and typically take γ1=0.5.VthIt is according to reality
Depending on testing needs, Vth=50.
Using γ-alignment technique, original image is carried out after brightness pretreatment, the pretreated image of brightness of gained is bright
Degree component V1(x, y) automatically becomes the average brightness value V of original imagea, and threshold value V with brightness pretreatment by itthCompare,
If threshold value V of no more than brightness pretreatmentth, then again brightness pretreatment is carried out using γ-alignment technique, then with brightness pretreatment
Threshold value VthCompare, iterative cycles, until the average brightness value V of original imageaThreshold value V more than brightness pretreatmentthTill,
Processed subsequently into next step.
If the average brightness value V of original imageaThreshold value V more than brightness pretreatmentth, then to former in HSV color space
Beginning image does not carry out brightness pretreatment.It should be noted that carry out brightness pretreatment subsequently all to need to carry out edge enhancing, because
This strengthens in step at follow-up edge, the average brightness value V of original imageaThreshold value V more than brightness pretreatmentthWhen, it is considered as
Carry out the brightness pretreatment that brightness keeps constant.
4. to brightness, pretreated image carries out edge enhancing, can more clearly show through the enhanced image in edge
Different types of object or the border of phenomenon, or the trace of linear image, in order to the identification of different types of object and its
The delineation of distribution, and then so that border is apparent from, image can be made to have more preferable visual effect.
In the present embodiment, using Laplacian detective operators, to brightness, pretreated image carries out edge enhancing,
Additive method can also be adopted, and to brightness, pretreated image carries out edge enhancing, and here does not limit.
Laplacian detective operators (Laplacian of Gaussian, LoG) are by gaussian filtering and Laplce
Detective operators are combined together, and carry out image border enhancing, commonly referred to Laplce's Gauss method.Using Laplacian
To brightness, pretreated image carries out edge enhancing, including herein below to detective operators：
First, using Gaussian function Gauss (x, y) to brightness pretreated luminance component image V1(x, y) is put down
Sliding filtering, i.e. gaussian filtering, then the luminance component image V after smoothing2(x, y) is as follows：
Wherein,Represent convolution algorithm；σ is space scale
The factor, in the present invention, σ value is taken as 2.25.
Secondly, using Laplce's detective operators to the luminance component image V after smoothing2(x, y) carries out edge enhancing, then
Edge enhanced image Vp(x, y), i.e. pretreated image Vp(x, y) is as follows：
Vp(x, y)=2[V2(x,y)] (7)
Wherein,2[] is binary second order gradient.
2) brightness according to pretreated image carries out region segmentation, and different special according to each region split
Point, selects corresponding mapping function, carries out corresponding adaption brightness enhancing respectively, to realize the adjustment of image overall brightness, its
Comprise the following steps：
1. the brightness according to pretreated image carries out image segmentation, so that the difference according to each region split
Feature carries out corresponding adaption brightness enhancing；
If through pretreated image VpThe mean flow rate of (x, y) is Va-p, by pretreated image VpThe institute of (x, y)
Pixel is had to be divided into two classes VlowAnd Vhigh：VlowRepresent the mean flow rate of low brightness pixel, VhighRepresent the averagely bright of high luminance pixel
Degree, respective expression formula is as follows：
Wherein,Sgn () is sign sign function.
By pretreated image VpThe pixel of (x, y) is divided into three types according to brightness：Brightness is [0, Vlow) interval
Pixel is pretreated image VpThe low-light (level) area pixel of (x, y)；Brightness is in [Vlow,Vhigh) interval pixel is pretreatment
Image V afterwardspThe normal region pixel of (x, y)；Brightness is in [Vhigh, 255] and interval pixel is pretreated image Vp(x,
Y) exposure area pixel.These three area pixel：Low-light (level) area pixel, normal region pixel and exposure area pixel draw
Point parameter is according to pretreated image Vp(x, y) self brightness feature is calculated, with these three parameters to image slices
Element is classified, so that subsequent treatment is more targeted, so that the present invention has stronger adaptability.
2. the different characteristics according to each region split, select corresponding mapping function, carry out respectively corresponding from
Adaptation brightness strengthens, to realize the adjustment of image overall brightness；
Pretreated image VpThe brightness histogram corresponding to three area pixel that the pixel of (x, y) is divided, selects
Following mapping function carries out brightness of image enhancing respectively, and its formula is as follows：
Wherein, VenhPixel brightness value at coordinate points (x, y) place, log1 after (x, y) expression brightness enhancing0() be with
10 is the logarithm at bottom：
Wherein, γ2It is also the parameter of γ-alignment technique, between (0,1).
3) using the feature of initial saturation and brightness flop, to segmented adaptive brightness, enhanced image carries out saturation
Degree enhancement process, to improve the visual effect of image enhaucament, and obtains final image.
Through the relatively pretreated image of the enhanced image of segmentation brightness (x, y) place pixel brightness flop Δ V
(x, y) is
Δ V (x, y)=[Venh(x,y)-Vp(x,y)]/255 (12)
Average staturation value S of the original image in HSV color spaceaFor
To segmentation brightness, enhanced image carries out saturation enhancement process, and its process is as follows：
Wherein, SenhThe pixel intensity value at (x, y) place for the image after (x, y) expression saturation enhancement process.In image
During saturation strengthens, brightness of image mapping transformation is combined with original image saturation, and gives mutual restriction, mutually accordingly
The weight coefficient of balance, so that image saturation strengthens has more preferable effect.It can be seen that saturation enhancing function is adopted
Coefficient etc. is also all determined by image own characteristic, so that the present invention has good reinforced effects and application model
Enclose.
Image after saturation enhancement process is converted back rgb color space from HSV color space, obtains final cromogram
Picture, its conversion formula is as follows：
Wherein, cos [] is cosine function, Renh(x, y), Genh(x, y) and Benh(x, y) is in rgb color space respectively
The enhanced red, green and blue component of (x, y) place pixel.The brightness of the image enhaucament that obtains using the present invention, highlight details, by
Strengthen in adding saturation, color is also more bright-coloured, is more suitable for the visually-perceptible of people.
In above-described embodiment, the image conversion between HSV color space and rgb color space can be enumerated using above-mentioned
Formula converted, it would however also be possible to employ other mark conversion formulas, and here do not limit.
In sum, the principle of the invention is as follows：First, low-light (level) in HSV color space and brightness disproportionation image are changed
To HSV color space, obtain original image, and original image is done with to carry out edge including brightness pretreatment and image enhanced pre-
Process.Then, the brightness according to pretreated image carries out region segmentation, and different special according to each region split
Point, selects corresponding mapping function, carries out corresponding adaption brightness enhancing respectively.Finally, using initial saturation and brightness
To segmented adaptive brightness, enhanced image carries out saturation enhancement process to the feature of change, and gained image is converted back to
Rgb color space, obtains final image.Combine because adaption brightness is strengthened to strengthen with adaptive saturation by the present invention,
Therefore all there are preferable reinforced effects for the image that Luminance Distribution is extremely uneven, brightness fluctuation is larger, so that image is had preferably
Visual effect.
In order to better illustrate effectiveness of the invention, illustrate below by Fig. 2～Fig. 4：
In result schematic diagram as shown in figs. 2 to 4, on the one hand can pass judgment on this from image angle process angle
Bright effectiveness, such as using picture appraisal standard：Image entropy is passed judgment on the image whether brightness of image processing through the present invention
It is evenly distributed, picture quality is good.
The definition of image entropy is：
Wherein, H (x) is image entropy, the probability that p (i) occurs in the whole gray value of image for image intensity value i.Image
Entropy is bigger, represents that image brightness distribution is more uniform, picture quality is better.As shown in table 1, mistake handled by method proposed by the present invention
Image entropy maximum, the details of image can preferably be strengthened so as to color is more bright-coloured using the present invention, be more suitable for regarding of people
Feel perception, improve visual quality of images that required for also allowing for extracting or region interested is further processed to image.
Table 1 image entropy
| Image entropy | a | b | c | d | e | f |
| Fig. 2 | 4.663 | 5.913 | 6.128 | 6.447 | 6.480 | 6.839 |
| Fig. 3 | 6.703 | 7.376 | 6.536 | 7.082 | 7.460 | 7.474 |
| Fig. 4 | 7.327 | 7.328 | 7.119 | 7.294 | 7.308 | 7.385 |
On the other hand can also prove to be respectively provided with more this paper algorithm complex more low enhancing degree from run time
Significantly reinforced effects.As shown in Table 2, run time under the conditions of VC6.0, OpenCV1.0 for each algorithm, by processing
Time it is also seen that run time of the present invention has a clear superiority, i.e. relatively low computation complexity.
Table 2 run time
| Run time/ms | b | c | d | e | f |
| Fig. 2 | 69 | 171 | 155 | 142 | 95 |
| Fig. 3 | 65 | 165 | 153 | 139 | 96 |
| Fig. 4 | 142 | 314 | 307 | 287 | 209 |
The above, the only present invention preferably specific embodiment, but protection scope of the present invention is not limited thereto,
Any those familiar with the art the invention discloses technical scope in, technology according to the present invention scheme and its
Inventive concept equivalent or change in addition, all should be included within the scope of the present invention.
Claims (3)
1. the method for adaptive image enhancement of a kind of low-light (level) or brightness disproportionation image, it comprises the following steps：
1) pretreatment is done to low-light (level) and brightness disproportionation image, pretreatment includes first doing brightness to low-light (level) and brightness disproportionation image
Pretreatment, and to brightness, pretreated image carries out edge enhancing, obtains pretreated image；
2) brightness according to pretreated image carries out region segmentation, and the different characteristics according to each region split,
Select corresponding mapping function, carry out corresponding adaption brightness enhancing respectively, to realize the adjustment of image overall brightness；
3) using the feature of initial saturation and brightness flop, to segmented adaptive brightness, enhanced image carries out saturation increasing
Strength is managed, and obtains final image,
Described step 1) comprise the following steps：
1. low-light (level) or brightness disproportionation image are transformed into HSV color space from rgb color space, to obtain HSV color space
In original image；
Choose low-light (level) or brightness disproportionation image in rgb color space, this picture size is M × N, and wherein, M is the height of this image
Degree, N is the width of this image；Any point coordinate in this image is (x, y), and x ∈ [0, M-1], y ∈ [0, N-1]；
For the pixel at any point (x, y) place, define R (x, y), G (x, y) and B (x, y) is this picture in rgb color space
The red component of element, green component and blue component；By the pixel of any point in low-light (level) or brightness disproportionation image from rgb color space
It is transformed into HSV color space, to realize for low-light (level) in rgb color space or brightness disproportionation image being transformed into HSV color space
In, obtain the original image in HSV color space, its each component of this pixel in HSV color space is：V (x, y) is brightness
Component, S (x, y) are saturation component and H (x, y) is chrominance component；
2. calculate the average brightness value V of the original image in HSV color spacea：
3. the threshold value of the average brightness value of original image and brightness pretreatment is compared, when the average brightness value of original image
It is not more than the threshold value of brightness pretreatment, then carry out brightness pretreatment, otherwise, then do not process；
Threshold value V according to brightness pretreatmentthAverage brightness value V with original imageaBetween magnitude relationship decide whether
Brightness pretreatment is carried out to the original image in HSV color space, its principle is as follows：
If the average brightness value V of original imageaIt is not more than threshold value V of brightness pretreatmentth, then adopt γ-alignment technique to HSV color
Original image in color space carries out brightness pretreatment；
Original image in HSV color space is carried out after the brightness pretreatment of brightness pretreatment gained using γ-alignment technique
Luminance component image V1(x,y)：
Wherein, γ1It is the parameter of γ-alignment technique, γ1Between (0,1), VthDepending on being needed according to experiment；
Using γ-alignment technique, original image is carried out after brightness pretreatment, bright according to the pretreated image of brightness of gained
Degree component V1(x, y) tries to achieve the average brightness value V of original imagea, and threshold value V with brightness pretreatment by itthCompare, if not
Threshold value V more than brightness pretreatmentth, then again brightness pretreatment, then the threshold with brightness pretreatment are carried out using γ-alignment technique
Value VthCompare, iterative cycles, until the average brightness value V of original imageaThreshold value V more than brightness pretreatmentthTill, then
Enter next step to be processed；If the average brightness value V of original imageaThreshold value V more than brightness pretreatmentth, then to HSV color
Original image in space does not carry out brightness pretreatment；
4. to brightness, pretreated image carries out edge enhancing, obtains pretreated image Vp(x,y).
2. the method for adaptive image enhancement of a kind of low-light (level) or brightness disproportionation image as claimed in claim 1, its feature exists
In：Described step 2) comprise the following steps：
1. the brightness according to pretreated image carries out region segmentation, so that the different characteristics according to each region split
Carry out corresponding adaption brightness enhancing；
If pretreated image VpThe mean flow rate of (x, y) is Va-p, by pretreated image VpThe all pixels of (x, y) are divided
For two classes VlowAnd Vhigh：VlowRepresent the mean flow rate of low brightness pixel, VhighRepresent the mean flow rate of high luminance pixel, each
Expression formula as follows：
Wherein,Sgn () is sign sign function；
By pretreated image VpThe pixel of (x, y) is divided into three types according to brightness difference：Brightness is [0, Vlow) interval
Pixel is pretreated image VpThe low-light (level) area pixel of (x, y)；Brightness is in [Vlow,Vhigh) interval pixel is pretreatment
Image V afterwardspThe normal region pixel of (x, y)；Brightness is in [Vhigh, 255] and interval pixel is pretreated image Vp(x,
Y) exposure area pixel；
2. the different characteristics according to each region split, selects corresponding mapping function, carries out corresponding self adaptation respectively
Brightness strengthens, to realize the adjustment of image overall brightness；
Pretreated image VpThe brightness histogram corresponding to three area pixel that the pixel of (x, y) is divided, selects to reflect as follows
Penetrate function and carry out brightness of image enhancing respectively, its formula is as follows：
Wherein, VenhPixel brightness value at coordinate points (x, y) place, log after (x, y) expression brightness enhancing10() is with 10 as bottom
Logarithm：
Wherein, γ2It is the parameter of γ-alignment technique, between (0,1).
3. the method for adaptive image enhancement of a kind of low-light (level) or brightness disproportionation image as claimed in claim 2, its feature exists
In：Described step 3) include herein below：
Through the relatively pretreated image of the enhanced image of segmentation brightness (x, y) place pixel brightness flop Δ V (x, y)
For
Δ V (x, y)=[Venh(x,y)-Vp(x,y)]/255
Average staturation value S of the original image in HSV color spaceaFor
To segmentation brightness, enhanced image carries out saturation enhancement process, and its process is as follows：
Wherein, SenhThe pixel intensity value at (x, y) place for the image after (x, y) expression saturation enhancement process；
Image after saturation enhancement process is converted back rgb color space from HSV color space, obtains final coloured image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410389246.2A CN104156921B (en) | 2014-08-08 | 2014-08-08 | Self-adaptive low-illuminance or non-uniform-brightness image enhancement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410389246.2A CN104156921B (en) | 2014-08-08 | 2014-08-08 | Self-adaptive low-illuminance or non-uniform-brightness image enhancement method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104156921A CN104156921A (en) | 2014-11-19 |
| CN104156921B true CN104156921B (en) | 2017-02-22 |
Family
ID=51882412
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410389246.2A Active CN104156921B (en) | 2014-08-08 | 2014-08-08 | Self-adaptive low-illuminance or non-uniform-brightness image enhancement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104156921B (en) |
Families Citing this family (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104517268B (en) * | 2014-12-24 | 2017-09-26 | 小米科技有限责任公司 | Adjust the method and device of brightness of image |
| CN104754184A (en) * | 2015-04-10 | 2015-07-01 | 四川理工学院 | Real-time monitoring video illumination compensation method and real-time monitoring video illumination compensation system |
| CN106441804B (en) * | 2015-08-04 | 2019-08-30 | 宁波舜宇光电信息有限公司 | Resolving power test method |
| CN105374018B (en) * | 2015-12-18 | 2018-10-19 | 厦门大学 | A method of region enhancing is carried out to image |
| CN105701780B (en) * | 2016-01-12 | 2018-09-28 | 中国科学院深圳先进技术研究院 | a kind of remote sensing image processing method and system |
| CN106023117B (en) * | 2016-06-01 | 2019-02-19 | 哈尔滨工业大学（威海） | Backlight image recovery method based on non-linear brightness lift scheme |
| CN106327437B (en) * | 2016-08-10 | 2019-04-05 | 大连海事大学 | A kind of color documents images bearing calibration and system |
| US10692194B2 (en) | 2016-09-30 | 2020-06-23 | Huawei Technologies Co., Ltd. | Method and terminal for displaying edge of rectangular frame |
| CN106504281B (en) * | 2016-12-02 | 2019-01-22 | 中国电子科技集团公司第四十四研究所 | Image quality enhancing and filtering method applied to cmos image sensor |
| CN107656454A (en) * | 2017-09-21 | 2018-02-02 | 深圳市晟达机械设计有限公司 | A kind of efficient cell monitoring management system |
| CN107644409B (en) * | 2017-09-28 | 2022-01-04 | 深圳Tcl新技术有限公司 | Image enhancement method, display device and computer-readable storage medium |
| CN108305227A (en) * | 2018-01-23 | 2018-07-20 | 中国航空工业集团公司洛阳电光设备研究所 | A kind of method that the image low brightness area that contrast is constant highlights |
| CN110136071B (en) * | 2018-02-02 | 2021-06-25 | 杭州海康威视数字技术股份有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN108364263B (en) * | 2018-02-05 | 2022-01-07 | 苏州沃科汽车科技有限公司 | Vehicle-mounted image processing method for standard definition input and high definition output |
| CN108470329A (en) * | 2018-02-12 | 2018-08-31 | 深圳市华星光电半导体显示技术有限公司 | A kind of image processing method and system |
| CN109064426B (en) * | 2018-07-26 | 2021-08-31 | 电子科技大学 | Method and device for suppressing glare in low-illumination image and enhancing image |
| CN109146811A (en) * | 2018-08-14 | 2019-01-04 | 长沙全度影像科技有限公司 | A kind of Adaptive contrast enhancement method of color image |
| CN109377462A (en) * | 2018-10-23 | 2019-02-22 | 上海鹰瞳医疗科技有限公司 | Method for processing fundus images and equipment |
| CN109493289B (en) * | 2018-10-26 | 2021-06-01 | 华侨大学 | Method for enhancing dual nonlinear images with brightness and saturation |
| CN109829859A (en) * | 2018-12-05 | 2019-05-31 | 平安科技（深圳）有限公司 | Image processing method and terminal device |
| CN109697706A (en) * | 2018-12-28 | 2019-04-30 | 哈尔滨工业大学 | A kind of self defined area Enhancement Method of soft image |
| CN111626965B (en) * | 2020-06-04 | 2021-03-16 | 成都星时代宇航科技有限公司 | Remote sensing image processing method, device, electronic equipment and storage medium |
| CN111931671A (en) * | 2020-08-17 | 2020-11-13 | 青岛北斗天地科技有限公司 | Face recognition method for illumination compensation in underground coal mine adverse light environment |
| WO2022087973A1 (en) * | 2020-10-29 | 2022-05-05 | Oppo广东移动通信有限公司 | Image processing method and apparatus, computer-readable medium, and electronic device |
| CN112541869A (en) * | 2020-12-07 | 2021-03-23 | 南京工程学院 | Retinex image defogging method based on matlab |
| CN114827482B (en) * | 2021-01-28 | 2023-11-03 | 抖音视界有限公司 | Image brightness adjusting method and device, electronic equipment and medium |
| CN113450272B (en) * | 2021-06-11 | 2024-04-16 | 广州方图科技有限公司 | Image enhancement method based on sinusoidal variation and application thereof |
| CN114049732B (en) * | 2021-09-29 | 2023-07-21 | 国网山东省电力公司郓城县供电公司 | Substation video monitoring method, system and storage medium |
| CN113822826B (en) * | 2021-11-25 | 2022-02-11 | 江苏游隼微电子有限公司 | Low-illumination image brightness enhancement method |
| CN114998159B (en) * | 2022-08-04 | 2022-10-28 | 邹城市天晖软件科技有限公司 | Design image self-adaptive enhancement method |
| CN115393228B (en) * | 2022-10-27 | 2023-03-24 | 摩尔线程智能科技(北京)有限责任公司 | Image processing method and device and graphic processing equipment |
| CN116528060B (en) * | 2023-07-04 | 2023-09-19 | 长春希达电子技术有限公司 | Dark light image enhancement device, method and device and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101340511A (en) * | 2008-08-07 | 2009-01-07 | 中兴通讯股份有限公司 | Adaptive video image enhancing method based on lightness detection |
| CN101742339A (en) * | 2010-01-14 | 2010-06-16 | 中山大学 | Method for enhancing color image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BRPI0302384B1 (en) * | 2002-07-20 | 2018-06-19 | Samsung Electronics Co., Ltd. | "METHOD FOR ADAPTABLE COLORING A COLOR, AND EQUIPMENT FOR ADAPTABLE COLORING AN IMAGE" |
-
2014
- 2014-08-08 CN CN201410389246.2A patent/CN104156921B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101340511A (en) * | 2008-08-07 | 2009-01-07 | 中兴通讯股份有限公司 | Adaptive video image enhancing method based on lightness detection |
| CN101742339A (en) * | 2010-01-14 | 2010-06-16 | 中山大学 | Method for enhancing color image |
Non-Patent Citations (1)
| Title |
|---|
| 彩色图像增强去噪研究;印文帅;《中国优秀硕士论文全文数据库 信息科技辑》;20120715;第44-45页第4.4节，图4.22 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104156921A (en) | 2014-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104156921B (en) | Self-adaptive low-illuminance or non-uniform-brightness image enhancement method | |
| Li et al. | Visual-salience-based tone mapping for high dynamic range images | |
| Lin et al. | Intensity and edge based adaptive unsharp masking filter for color image enhancement | |
| Huang et al. | Efficient contrast enhancement using adaptive gamma correction with weighting distribution | |
| Ma et al. | An effective fusion defogging approach for single sea fog image | |
| CN104574293B (en) | Multiple dimensioned Retinex image sharpenings algorithm based on bounded computing | |
| Vishwakarma et al. | Color image enhancement techniques: a critical review | |
| CN109919859B (en) | Outdoor scene image defogging enhancement method, computing device and storage medium thereof | |
| Park et al. | Contrast enhancement for low-light image enhancement: A survey | |
| Sun et al. | Brightness preserving image enhancement based on a gradient and intensity histogram | |
| CN105931206A (en) | Method for enhancing sharpness of color image with color constancy | |
| Jiang et al. | Color image enhancement with brightness preservation using a histogram specification approach | |
| CN111968065A (en) | Self-adaptive enhancement method for image with uneven brightness | |
| Priyanka et al. | Low-light image enhancement by principal component analysis | |
| CN105427255A (en) | GRHP based unmanned plane infrared image detail enhancement method | |
| Parihar et al. | A comprehensive analysis of fusion-based image enhancement techniques | |
| Zhang et al. | Image dehazing based on dark channel prior and brightness enhancement for agricultural remote sensing images from consumer-grade cameras | |
| Wen et al. | Autonomous robot navigation using Retinex algorithm for multiscale image adaptability in low-light environment | |
| Anjana et al. | Color image enhancement using edge based histogram equalization | |
| CN116309146A (en) | Retinex-based edge-preserving color low-illumination image enhancement method | |
| Parihar | Histogram modification and DCT based contrast enhancement | |
| Lavania et al. | A comparative study of image enhancement using histogram approach | |
| David | Low illumination image enhancement algorithm using iterative recursive filter and visual gamma transformation function | |
| Kumar et al. | 3D color channel based adaptive contrast enhancement using compensated histogram system | |
| Tang et al. | Sky-preserved image dehazing and enhancement for outdoor scenes |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |