CN105096278A - Image enhancement method based on illumination adjustment and equipment thereof - Google Patents

Abstract

The invention provides an adaptive Gamma enhancement method based on a Retinex theory. The Retinex theory is used for separating the brightness component and the reflection component of an image. Adaptive Gamma correction is performed on the brightness component. Finally the detail and the color of the image are restored through the reflection component. The adaptive Gamma enhancement method settles problems of poor timeliness and low processing effect of an existing Retinex algorithm. The processed heterogeneous illumination image has an optimal contrast, an optimal visibility, an optimal naturalness and an optimal timeliness.

Description

Based on image enchancing method and the equipment of illumination adjustment

Technical field

The present invention relates to image processing field, particularly based on image enchancing method and the equipment of illumination adjustment.

Background technology

Constantly upgrade along with mobile device performance and popularize, in daily life, people more and more get used to taking pictures to things interested or making a video recording anywhere or anytime, but the generation of these images or video is often under nonrestrictive condition, this just causes there is the even situation of uneven illumination in image.Process for inhomogeneous illumination image has become image processing field problem in the urgent need to address.

In order to solve the even problem of uneven illumination, scientific research personnel has carried out large quantifier elimination.Retinex (RetinaandCortex) theory is widely used in the enhancing of the even image of uneven illumination.This theory hypothesis image is combined into by luminance component and reflecting component, and two components can separating treatment.Early stage Retinex algorithm, while compensating light photograph, has suppressed the dynamic range of image, and there will be halation and cross-color phenomenon.In order to solve halation and color distortion problem, adopt the method for multiple dimensioned Retinex tone mapping to strengthen inhomogeneous illumination image, but enhancing image seem natural not.In conjunction with Retinex principle and double-log luminance filter, improve visibility and the naturality of image further, but its computation complexity is very high, greatly consuming time, brightness change greatly.

Summary of the invention

The present invention solves existing algorithm poor real and strengthens the undesirable problem of effect, propose a kind of self-adaptation Gamma based on Retinex theory and strengthen algorithm, utilize luminance component and the reflecting component of the theoretical separate picture of Retinex, self-adaptation Gamma correction is carried out to luminance component, finally utilizes details and the color of reflecting component Recovery image.The results show, it is higher that algorithm of carrying has image definition, strengthens effect more natural, the ageing advantage such as better.

Accompanying drawing illustrates:

Fig. 1 is the image enchancing method based on illumination adjustment;

Fig. 2 is the enhancing result of image Boy;

Fig. 3 is the enhancing result of image Museum;

Fig. 4 is the enhancing result of image Cockpit;

Fig. 5 is the enhancing result of image Girl;

Fig. 6 is the enhancing result of image Hall;

Fig. 7 is the enhancing result of image Building;

Fig. 8 is application image Enhancement Method in ISP chip.

Technical scheme

As shown in Figure 1, Fig. 1 gives the algorithm for image enhancement process flow diagram based on illumination adjustment, herein by for several dissimilar degraded image, passes through proposed algorithm for image enhancement and processes, and contrast from different algorithms respectively, verify herein put forward performance and the versatility of algorithm.

Several step is divided into the algorithm for image enhancement adjusted based on illumination:

Step one: obtain observed image I (x, y);

Step 2: decompose the image observed, becomes reflecting component and luminance component by picture breakdown;

Retinex theory thinks that image is made up of luminance component and reflecting component.Suppose that each passage of RGB image I (x, y) has identical brightness, that is:

I ^c(x,y)＝R ^c(x,y)·L(x,y),c∈{r,g,b}(1)

R in formula ^c(x, y) represents the reflecting component of each passage, and L (x, y) represents the luminance component of image.Usually, using the illumination V (x, y) of the maximal value of RGB passage as eye-observation.

$V(x,y)=\underset{c\∈\{r,g,b\}}{\max }{I}^c(x,y)---\left(2\right)$

Theoretical according to Retinex, with 2D Gaussian filter G (x, y), the luminance component L (x, y) that convolution can obtain image is carried out to illumination image V (x, y).

L(x,y)＝G(x,y)*V(x,y)(3)

Thus, reflecting component R ^c(x, y) can be separated.

R ^c(x,y)＝I ^c(x,y)/L(x,y),c∈{r,g,b}(4)

Reflecting component mainly contains the radio-frequency component of image, comprises edge and details.

Step 3: Gamma conversion is carried out to luminance component;

After acquisition luminance component image, self-adaptation Gamma correction is carried out to it.

L _en(x,y)＝L(x,y) ^γ(x,y)(5)

$\mathrm{\γ}\left(l\right)=1-\underset{v=0}{\overset{l}{\Σ}}\[P_{\mathrm{\ω}}\left(v\right)/s_p\]---\left(6\right)$

$s_p=\underset{l=0}{\overset{l_{max}}{\Σ}}{P}_{\mathrm{\ω}}\left(l\right)---\left(7\right)$

L _en(x, y) represents the luminance component after strengthening, and γ (x, y) represents Gamma correction coefficient matrix, P _ωl () is for corresponding to the weights distribution function of each brightness value:

$P_{\mathrm{\ω}}\left(l\right)=\frac{P\left(l\right)-p_{\min }}{p_{max}-p_{\min }}---\left(8\right)$

In formula, P (l) is the probability density function of luminance component, p _maxfor the maximal value of P (l), p _minfor the minimum value of P (l), P (l) tries to achieve by following formula:

P(l)＝n _l/n _p(9)

In formula, n _lpixel count contained by the brightness of correspondence, n _pfor the sum of all pixels that luminance component comprises.

Step 4: by the luminance component synthesis after reflecting component and conversion;

Merge L _en(x, y) and R ^c(x, y) can obtain final enhancing image I _en(x, y).

$I_{en}^c(x,y)=R^c(x,y)\·L_{en}(x,y),c\∈\{r,g,b\}---\left(10\right)$

Step 5: form the image strengthened.

The image quality evaluation strengthened:

For evaluating the performance of algorithm herein, choosing advanced Retinex inhomogeneous illumination algorithm for image enhancement, carrying out subjective assessment, objective evaluation and ageing comparison.

1. subjective assessment

Fig. 2 ~ 7 are the inhomogeneous illumination processing result image of different scene.The former figure of Fig. 2 is that fine day is outdoor, and the face of boy is in shade, face's low visibility.The former figure of Fig. 3 is that indoor weak light shines image, and the scene scenery after showcase is unintelligible.Glass reflecting is had, the things low visibility in aircraft cabin in the former figure of Fig. 4.In the former figure of Fig. 5, because illumination is blocked, right side seems very dark.The former figure of Fig. 6 is indoor inhomogeneous illumination image.Fig. 7 is the result of outdoor cloudy day building photo, and in former figure red block, the result of topography lists in the below of corresponding enhancing image.

Document 1 (Zhang Shangwei, Zeng Ping, Luo Xuemei, Deng. there is the multiple dimensioned Retinex tone-mapping algorithm [J] of details compensation and color recovery. XI AN JIAOTONG UNIVERSITY Subject Index, 2012,46 (4): 32-37) result of algorithm strengthens effect not obviously, and the dynamic range of image is compressed.Document 2 (WangS, ZhengJ, HuH, etal.Naturalnesspreservedenhancementalgorithmfornon-unif ormilluminationimages [J] .IEEETransactionsonImageProcessing, 2013,22 (9): 3538 – 3548.) result of algorithm makes image be largely increased in the sharpness of the overall situation and local, but the dynamic range of image is also pressed.Algorithm of the present invention can not only significantly improve the sharpness of image overall and regional area, and makes image have good dynamic range and subjective natural sense.

2. objective evaluation

In order to carry out objective appraisal to process image, introduce EBCM (edgebasedcontrastmeasure) herein, the evaluation indexes such as VE (visibleedges), NIQE (naturalnessimagequalityevaluator).Wherein EBCM is used for the contrast of evaluation map picture, and this parameter values is larger, then the contrast of image is higher.VE is for evaluating the ratio strengthening image and improve relative to the visibility of former figure, and its value is larger, then in image, things visibility is higher.NIQE is used for the naturality of evaluation map picture, and its value is less, and the naturality of key diagram picture is better.

Table 1 ~ 3 list the evaluation index result of image in Fig. 2 ~ 7 respectively.As can be seen from the table, after each algorithm process, image is obtained for relative to former figure and significantly improves in contrast, visibility, naturality three.Method herein obviously has better performance than algorithm in document [1] and document [2]: in contrast, on average improve 49.7% and 10.8% respectively; In visibility, on average improve 102.4% and 38.5% respectively; In naturality, improve 31% and 16.4% respectively.

The EBCM result of each algorithm of table 1

The VE result of each algorithm of table 2

The NIQE result of each algorithm of table 3

3. ageing comparison

Ageing in order to more each algorithm, it is the on average consuming time of 2000 × 1312 images that table 4 shows various algorithm to size, and these data are the test results utilizing Matlab2014 on hardware parameter 3.3GHzCPU, 4GBRAM computer.Obviously, herein institute's extracting method is ageing best, be only document [6] 11% and document [7] consuming time consuming time 2%.

Process 2000 × 1312 image of each algorithm of table 4 on average consuming time

Can observe from experimental result, self-adaptation Gamma proposed by the invention strengthens algorithm can become luminance picture and albedo image by a width observation RGB separation of images.Because the brightness only for image processes, do not introduce colouring information.Therefore isolated luminance picture and albedo image are gray level image, and wherein luminance picture has reacted the monochrome information of surrounding environment well, has the character of space smoothing simultaneously.Albedo image then remains edge and the detailed information of image self, and these features demonstrate the new objective function that this title of the song proposes well, i.e. the rationality of formula (10).And under rgb space, luminance picture and albedo image, except having the characteristic of the result of calculation under HSV space, also show colouring information.As can be seen from result, corrected adjust to illumination the enhancing image obtained in conjunction with reflectivity again by Gamma, except strengthening detail section, while promoting the brightness of dark area, subjective vision effect also has to be improved definitely.

The algorithm that the present invention proposes can be applied on the images such as TV, mobile phone, the video camera common apparatus relevant with video.The figure that can be applied to specific occasion (medical science, military affairs, public safety etc.) strengthens.This section introduces the application of algorithm on high-definition camera.

Usually, in order to improve graphics process performance, high-definition camera is all integrated with image-signal processor ISP (ImageSingalProcessor), signal for exporting imageing sensor (CCD or CMOS) carries out post-processed (as: accumulation of 3D noise reduction, frame, high light suppress), and then strengthen the quality of institute's output image, meet the demand of application-specific.By algorithm integration herein in ISP chip, can realize the image enhaucament under inhomogeneous illumination condition.System composition as shown in Figure 7.

The present invention proposes a kind of self-adaptation Gamma Enhancement Method based on Retinex theory, utilize luminance component and the reflecting component of the theoretical separate picture of Retinex, self-adaptation Gamma correction is carried out to luminance component, finally utilizes details and the color of reflecting component Recovery image.Herein algorithm solves existing Retinex algorithm poor in timeliness and the undesirable problem for the treatment of effect, and the inhomogeneous illumination image of this algorithm process has best contrast, visibility, naturality and ageing.Herein algorithm to hardware without particular/special requirement, portable to various TV, mobile phone, video camera or other have in the electronic product of image displaying function.

Claims (2)

1., based on an image enchancing method for illumination adjustment, it is characterized in that, this method is divided into several step:

Step one: obtain observed image I (x, y);

Step 2: decompose the image observed, becomes reflecting component and luminance component by picture breakdown;

Retinex theory thinks that image is made up of luminance component and reflecting component, supposes that each passage of RGB image I (x, y) has identical brightness, that is:

I ^c(x,y)＝R ^c(x,y)·L(x,y),c∈{r,g,b}(1)

R in formula ^c(x, y) represents the reflecting component of each passage, and L (x, y) represents the luminance component of image.Usually, using the illumination V (x, y) of the maximal value of RGB passage as eye-observation,

$V(x,y)=\underset{c\∈\{r,g,b\}}{max}{I}^c(x,y)---\left(2\right)$

Theoretical according to Retinex, with 2D Gaussian filter G (x, y), the luminance component L (x, y) that convolution can obtain image is carried out to illumination image V (x, y),

L(x,y)＝G(x,y)*V(x,y)(3)

Thus, reflecting component R ^c(x, y) can be separated,

R ^c(x,y)＝I ^c(x,y)/L(x,y),c∈{r,g,b}(4)

Reflecting component mainly contains the radio-frequency component of image, comprises edge and details,

Step 3: after acquisition luminance component image, self-adaptation Gamma correction is carried out to it;

L _en(x,y)＝L(x,y) ^γ(x,y)(5)

$\mathrm{\γ}\left(l\right)=1-\underset{v=0}{\overset{l}{\Σ}}\[P_{\mathrm{\ω}}\left(v\right)/s_p\]---\left(6\right)$

$s_p=\underset{l=0}{\overset{l_{max}}{\Σ}}{P}_{\mathrm{\ω}}\left(l\right)---\left(7\right)$

L _en(x, y) represents the luminance component after strengthening, and γ (x, y) represents Gamma correction coefficient matrix, P _ωl () is for corresponding to the weights distribution function of each brightness value:

$P_{\mathrm{\ω}}\left(l\right)=\frac{P\left(l\right)-p_{\min }}{p_{max}-p_{\min }}---\left(8\right)$

In formula, P (l) is the probability density function of luminance component, p _maxfor the maximal value of P (l), p _minfor the minimum value of P (l), P (l) tries to achieve by following formula:

P(l)＝n _l/n _p(9)

In formula, n _lpixel count contained by the brightness of correspondence, n _pfor the sum of all pixels that luminance component comprises;

Step 4: merge L _en(x, y) and R ^c(x, y) can obtain final enhancing image I _en(x, y).

$I_{en}^c(x,y)=R^c(x,y)\·L_{en}(x,y),c\∈\{r,g,b\}---\left(10\right)$

2. a high-performance image signal processor ISP, is characterized in that, employs method according to claim 1.