CN102938837B - A kind of tone mapping method keeping full sub-model based on edge - Google Patents
A kind of tone mapping method keeping full sub-model based on edge Download PDFInfo
- Publication number
- CN102938837B CN102938837B CN201210350895.2A CN201210350895A CN102938837B CN 102938837 B CN102938837 B CN 102938837B CN 201210350895 A CN201210350895 A CN 201210350895A CN 102938837 B CN102938837 B CN 102938837B
- Authority
- CN
- China
- Prior art keywords
- image
- light intensity
- log
- value
- edge
- 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 present invention discloses a kind of tone mapping method keeping full sub-model based on edge, comprising: input a width high-dynamics image; The high-dynamics image of input is carried out light intensity reconstruct; Obtain the logarithmic value of the light intensity value of new construction; Utilize edge to keep full variation wave filter that the light intensity logarithmic value obtained is carried out filtering and obtain image basal layer; Light intensity logarithmic value subtracted image basal layer is utilized to obtain the levels of detail of image; Calculate compression factor Summing Factor image output light intensity; A new scale factor M is obtained divided by input picture light intensity by light intensity after compression. Scale factor M is acted on respectively on red channel, green channel, each passage of blue channel, obtain the image after compressing color; Carry out gamma correction obtain low dynamic image by obtaining compressing the image after color. Keep the tone mapping method of Total Variation can obtain the very high low dynamic image of quality based on edge, halation phenomenon can not be produced.
Description
Technical field
The present invention relates to technical field of image processing, it is specifically related to keep based on edge the tone mapping method of full sub-model.
Background technology
High-dynamics image (HighDynamicRangeImage, HDRImage) can support very big range of light intensities, it is possible to effectively stores the intensity information of real world, presents the image of superior quality to people. But, traditional display equipment and printer can not show these high-dynamics images. Because the contrast gradient that they can show is far smaller than high-dynamics image. In order to address this problem, high-dynamics image dynamicrange can be compressed at present and make it adapt to low dynamic image display. The method that high-dynamics image is converted into low dynamic image (LowDynamicRangeImage, LDRImage) as this kind is exactly tone mapping method.
Now, a large amount of tone mapping methods is suggested. The compression of the simplest contrast gradient is exactly that original image is multiplied by a scale factor C, C < 1. This can lose details and the texture of shade or high-brightness region in image scene. This just requires that we find one method and compress contrast gradient as far as possible, keeps the texture in image and details simultaneously. A lot of tone mapping method has used picture breakdown technology. Piece image is decomposed into basal layer and levels of detail. Basal layer has high-contrast to be needed to be compressed, and levels of detail has texture clearly to be needed to be retained. Only being compressed by basal layer, levels of detail remains unchanged, and then by the basal layer synthesis after levels of detail and compression, just obtains low dynamic image. Basal layer can be obtained by edge preserving smoothing method, and levels of detail subtracts basal layer by input picture and obtains. DurandandDorsey describes a kind of tone mapping method based on quick bilateral filtering at document " Fastbilateralfilteringforthedisplayofhigh-dynamic-rangei mages " (2002, pp.257-266). This kind utilizes quick two-sided filter that image is carried out multi-resolution decomposition, obtains basal layer and levels of detail. Quick two-sided filter effectively smoothly small details can keep strong edge simultaneously. But, increasing smoothness, image border will be fuzzy. This makes the LDRimage obtained by the method halation phenomenon occur.
Tone mapping method based on quick bilateral filtering effectively smoothly small details can keep strong edge simultaneously. But, increasing smoothness, image border will be fuzzy. This makes the LDRimage obtained by the method halation phenomenon occur.
Summary of the invention
Instant invention overcomes the shortcoming that the tone mapping method based on quick bilateral filtering can produce halation phenomenon, it is proposed that a kind of tone mapping method keeping Total Variation based on edge.
The present invention provides a kind of tone mapping method keeping full sub-model based on edge, comprising:
Input a width high-dynamics image;
The high-dynamics image of input is carried out light intensity reconstruct;
Obtain the logarithmic value of the light intensity value of new construction;
Utilize edge to keep full variation wave filter that the logarithmic value of the light intensity value obtained is carried out filtering and obtain image basal layer;
Input picture subtracted image basal layer is utilized to obtain the levels of detail of image;
Calculate compression factor Summing Factor image output light intensity;
A new scale factor M is obtained divided by light intensity value intensity with image output light intensity. Scale factor M is acted on respectively on red channel, green channel, each passage of blue channel, obtain the image after compressing color;
Carry out gamma correction obtain low dynamic image by obtaining compressing the image after color.
The described high-dynamics image to input carries out light intensity reconstruct and comprises:
Channel value according to red channel R, green channel G, blue channel B, utilizes formula intensity=0.299*R+0.587*G+0.114*B, re-constructs out light intensity value intensity.
Described calculating compression factor Summing Factor image output light intensity comprises:
Calculate the compression factor factor: utilize formula compressfactor=log (outputrange)/(max (log_base)-min (log_base)) * 1.5, wherein outputrange is the dynamic range of images exported, max (log_base) is the maximum value getting all pixel values in image basal layer log_base, min (log_base) is the minimum value getting all pixel values in image basal layer log_base, and 1.5 is empirical value;
Computed image output light intensity: utilize formula output_intensity=exp (compressfactor*log_base+log_detail).
A new scale factor M is obtained divided by light intensity value intensity with image output light intensity. Scale factor M is acted on respectively on red channel, green channel, each passage of blue channel, obtains the image after compressing color and comprise:
Utilize formula
Wherein, output_intensity is image output light intensity;
Obtain R, G, B tri-color channel compression factor factors; Value after R, G, B compression is easy to be calculated;
Wherein R_Output, G_Output, B_Output are respectively compressed images R, G, B color-values, R_Input, G_Input, B_Input are respectively input picture R, G, B color-values, the calculating done is all on the logarithm of image light intensity, whole image range, with regard to the contrast gradient of correspondence image, is obtained unified process by the difference of each pixel value of this sampled images simultaneously. Above technology is it may be seen that the tone mapping method based on edge maintenance Total Variation can obtain the very high low dynamic image of quality, and the process effect of color and contrast gradient is very good, and can not produce halation phenomenon.
Accompanying drawing explanation
In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, it is briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.
Fig. 1 is the tone mapping method schema keeping Total Variation based on edge in the embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only the present invention's part embodiment, instead of whole embodiments. Based on the embodiment in the present invention, those of ordinary skill in the art, not making other embodiments all obtained under creative work prerequisite, belong to the scope of protection of the invention.
Instant invention overcomes the shortcoming that the tone mapping method based on quick bilateral filtering can produce halation phenomenon, it is proposed that a kind of tone mapping method keeping Total Variation based on edge. Total variation (TotalVariation) regularization model was proposed by Rudin, Osher and Fatemi in 1992, with the noise removed in image. This model has two base attributes: one is the edge that can keep image, and it is very good to process effect under specified conditions; Another is that the variable quantity of image light intensity and image detail yardstick are inversely proportional to. More succinct says, this model can effectively keep edge while smooth grain. This can be realized by minimumization equation.
Wherein Section 1 is fidelity item, ensures that the image u after smoothly retains the main feature observing image f; Section 2 is canonical item, is reached the object of smoothed image by the total variation of minimumization u. �� > 0 is scaled factor, plays the effect of balance fidelity item and canonical item. The value of �� is more big, and smooth effect is more obvious. K is the linear operator that determinacy is degenerated. The Euler-Lagrange equation that formula (1) is corresponding is
Wherein K*It it is the conjugation operator of K. Formula (1) and formula (2) have identical solution, and can in the hope of unique solution.
Adopt steepest descent to solve formula (1), add suitable first, boundary values condition, be then constructed as follows reaction-diffusion equation,
Wherein,Being the border of ��, n is borderOn the outer method vector of unit. Method of finite difference is adopted to solve formula (3).
Utilizing edge recited above to keep operator, two yardsticks that we are easy to realize image decompose. Resolve into the levels of detail of piecewise smooth, the basal layer of high-contrast and low contrast, small scale texture. Basal layer keeps total variation wave filter to obtain by edge, subtracts basal layer with input picture and just obtains levels of detail. Saying more specifically, g represents a width input picture and is used to two yardsticks decomposition, fTVFor edge keeps function of total variation, layer based on b, d is levels of detail, and said process is expressed as
B=fTV(g),(4)
D=g-b. (5)
The object of tone mapping be a width high-dynamics image is converted into a width can conventional display apparatus display low dynamic image. In this process, it is necessary to reduce contrast gradient, reach the dynamicrange that conventional display apparatus can show. Meanwhile, in order to obtain a high-quality low dynamic image, it is necessary to the details in the former high-dynamics image of reservation as much as possible. Basal layer comprises high-contrast information, and levels of detail comprises detail textures information. So only basal layer need to be carried out yardstick compression, levels of detail remains unchanged, so that it may to reach the object reducing contrast gradient and retaining details simultaneously. Finally, basal layer after compression and levels of detail are synthesized new images, is required LDR image. It is expressed as formula form as follows
U=C*b+d, (6)
Wherein, C is the compression factor factor, and C < 1, u is the low dynamic image after synthesis. Formula (5) is brought into formula (6) obtain
U=g-(1-C) * b, (7)
By formula (7), more significantly seeing, only the big scale of image being compressed, other small scale details are retained. Luminance brightness compression degree can be regulated, to meet the vision effect that the mankind want by changing scaled factor C. Formula (4) is brought into the tone mapping method that formula (7) just obtains this patent and proposes
U=g-(1-C) * fTV(g).(8)
For the process of color of image, first we calculate image output light intensity, with image output light intensity divided by light intensity value intensity, obtains a new scale factor M like this. Scale factor M is acted on respectively in R, G, channel B, just obtain the image after compressing color. Saying more specifically, I_Input is light intensity value intensity, I_Output is image output light intensity. Then
This results in R, G, B tri-color channel compression factor factors. Value after R, G, B compression is easy to be calculated. As follows
Wherein R_Output, G_Output, B_Output are respectively compressed images R, G, B color-values, and R_Input, G_Input, B_Input are respectively input picture R, G, B color-values. The calculating that we do is all on the logarithm of image light intensity, and whole image range, with regard to the contrast gradient of correspondence image, can be obtained unified process by the difference of each pixel value of this sampled images simultaneously.
Finally image is carried out gammacorrection. The method can obtain the very high low dynamic image of quality, and the process effect of color and contrast gradient is very good, and can not produce halation phenomenon.
Illustrating the tone mapping method keeping full sub-model based on edge in the embodiment of the present invention, concrete steps are as follows:
S101: input a width high-dynamics image.
S102: carry out light intensity (intensity) reconstruct. Channel value according to red channel (R), green channel (G), blue channel (B), utilizes formula intensity=0.299*R+0.587*G+0.114*B, re-constructs out light intensity value.
S103: the logarithmic value (log_intensity) obtaining the light intensity value of new construction.
S104: image filtering. Utilize the logarithmic value (log_intensity) of the light intensity value that edge keeps full variation wave filter step 3 to be obtained to carry out filtering and obtain image basal layer (log_base).
S105: picture breakdown. Utilize input picture to subtract the image basal layer (log_base) of step S104, obtain the levels of detail (log_detail) of image. So just a width high-dynamics image is decomposed into image basal layer and image detail layer.
S106: calculate compression factor factor compressfactor. Utilize formula compressfactor=log (outputrange)/(max (log_base)-min (log_base)) * 1.5, wherein outputrange is the dynamic range of images exported, max (log_base) is the maximum value getting all pixel values in image basal layer log_base, min (log_base) is the minimum value getting all pixel values in image basal layer log_base, and 1.5 is empirical value.
S107: computed image output light intensity output_intensity. Utilize formula output_intensity=exp (compressfactor*log_base+log_detail).
S108: color treatments. Calculate image output light intensity output_intensity by step S107, with image output light intensity divided by light intensity value intensity, obtain a new scale factor M like this. Scale factor M is acted on respectively on red channel (R), green channel (G), blue channel (B) each passage, just obtain the image after compressing color. See that formula is as follows:
Wherein, output_intensity is image output light intensity;
This results in R, G, B tri-color channel compression factor factors. Value after R, G, B compression is easy to be calculated. As follows
Wherein R_Output, G_Output, B_Output are respectively compressed images R, G, B color-values, and R_Input, G_Input, B_Input are respectively input picture R, G, B color-values. The calculating that we do is all on the logarithm of image light intensity, and whole image range, with regard to the contrast gradient of correspondence image, can be obtained unified process by the difference of each pixel value of this sampled images simultaneously.
S109:gamma corrects. So just obtain low dynamic image.
To sum up, keeping the tone mapping method of Total Variation can obtain the very high low dynamic image of quality based on edge, the process effect of color and contrast gradient is very good, and can not produce halation phenomenon.
The all or part of step that one of ordinary skill in the art will appreciate that in the various methods of above-described embodiment can be completed by the hardware that program carrys out instruction relevant, this program can be stored in a computer-readable recording medium, storage media can comprise: read-only storage (ROM, ReadOnlyMemory), random access memory (RAM, RandomAccessMemory), disk or CD etc.
What the embodiment of the present invention provided above keeps the tone mapping method of Total Variation based on edge, it is described in detail, apply specific case herein the principle of the present invention and enforcement mode to have been set forth, illustrating just for helping the method understanding the present invention and core concept thereof of above embodiment; Meanwhile, for one of ordinary skill in the art, according to the thought of the present invention, all will change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (3)
1. one kind keeps the tone mapping method of full sub-model based on edge, it is characterised in that, comprising:
Input a width high-dynamics image;
Channel value according to red channel R, green channel G, blue channel B, utilizes formula intensity=0.299*R+0.587*G+0.114*B, re-constructs out light intensity value intensity;
Obtain the logarithmic value of the light intensity value of new construction;
Utilize edge to keep full variation wave filter that the logarithmic value of the light intensity value obtained is carried out filtering and obtain image basal layer;
Input picture subtracted image basal layer is utilized to obtain the levels of detail of image;
Calculate compression factor Summing Factor image output light intensity;
Obtain a new scale factor M with image output light intensity divided by light intensity value intensity, scale factor M is acted on respectively on red channel, green channel, each passage of blue channel, obtain the image after compressing color;
Carry out gamma correction obtain low dynamic image by obtaining compressing the image after color;
Wherein, the described step utilizing edge to keep full variation wave filter that the logarithmic value of the light intensity value obtained carries out filtering acquisition image basal layer comprises: pass through log_base=fTV(log_intensity) logarithmic value of the light intensity value of input picture being carried out two yardstick decomposition, wherein, log_intensity is the logarithmic value of the light intensity value of a width input picture, fTVFor edge keeps function of total variation, layer based on log_base.
2. the tone mapping method keeping full sub-model based on edge as claimed in claim 1, it is characterised in that, described calculating compression factor Summing Factor image output light intensity comprises:
Calculate the compression factor factor: utilize formula compressfactor=log (outputrange)/(max (log_base)-min (log_base)) * 1.5, wherein outputrange is the dynamic range of images exported, max (log_base) is the maximum value getting all pixel values in image basal layer log_base, min (log_base) is the minimum value getting all pixel values in image basal layer log_base, and 1.5 is empirical value;
Computed image output light intensity: utilize formula output_intensity=exp (compressfactor*log_base+log_detail), wherein, log_detail is image detail layer.
3. the tone mapping method keeping full sub-model based on edge as claimed in claim 2, it is characterized in that, a new scale factor M is obtained divided by light intensity value intensity with image output light intensity, scale factor M is acted on respectively on red channel, green channel, each passage of blue channel, obtains the image after compressing color and comprise:
Utilize formula
Wherein, output_intensity is image output light intensity;
Obtain R, G, B tri-color channel compression factor factors; By following formulae discovery go out R, G, B compress after value:
Wherein R_Output, G_Output, B_Output are respectively compressed images R, G, B color-values, the calculating done is all on the logarithm of image light intensity, the contrast gradient of the different correspondence image of each pixel value of image, obtains unified process simultaneously to whole image range.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210350895.2A CN102938837B (en) | 2012-09-19 | 2012-09-19 | A kind of tone mapping method keeping full sub-model based on edge |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210350895.2A CN102938837B (en) | 2012-09-19 | 2012-09-19 | A kind of tone mapping method keeping full sub-model based on edge |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102938837A CN102938837A (en) | 2013-02-20 |
CN102938837B true CN102938837B (en) | 2016-06-01 |
Family
ID=47697703
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210350895.2A Active CN102938837B (en) | 2012-09-19 | 2012-09-19 | A kind of tone mapping method keeping full sub-model based on edge |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102938837B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103702116B (en) * | 2013-12-06 | 2017-07-14 | 浙江宇视科技有限公司 | A kind of dynamic range compression method and apparatus of image |
WO2015192368A1 (en) | 2014-06-20 | 2015-12-23 | 深圳市大疆创新科技有限公司 | Hdri generating method and apparatus |
CN104065939B (en) * | 2014-06-20 | 2017-03-22 | 深圳市大疆创新科技有限公司 | HDRI generating method and device |
WO2016056787A1 (en) * | 2014-10-06 | 2016-04-14 | Samsung Electronics Co., Ltd. | Display device and method of controlling the same |
WO2018068744A1 (en) * | 2016-10-14 | 2018-04-19 | Mediatek Inc. | Method and apparatus of smoothing filter for ringing artefact removal |
CN106847149B (en) * | 2016-12-29 | 2020-11-13 | 武汉华星光电技术有限公司 | Tone mapping and displaying method for high dynamic contrast image |
CN110493584B (en) * | 2019-07-05 | 2022-04-29 | 湖北工程学院 | High dynamic range scene visualization method and device and storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102388612A (en) * | 2009-03-13 | 2012-03-21 | 杜比实验室特许公司 | Layered compression of high dynamic range, visual dynamic range, and wide color gamut video |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9024951B2 (en) * | 2011-02-16 | 2015-05-05 | Apple Inc. | Devices and methods for obtaining high-local-contrast image data |
-
2012
- 2012-09-19 CN CN201210350895.2A patent/CN102938837B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102388612A (en) * | 2009-03-13 | 2012-03-21 | 杜比实验室特许公司 | Layered compression of high dynamic range, visual dynamic range, and wide color gamut video |
Non-Patent Citations (2)
Title |
---|
Fast Bilateral Filtering for the Display of High-Dynamic-Range Images;Durand and Dorsey;《ACM Transactions on Graphics》;20021231;第257-266页 * |
基于概率模型的高动态范围图像色调映射;宋明黎;《软件学报》;20090331;第734-742页 * |
Also Published As
Publication number | Publication date |
---|---|
CN102938837A (en) | 2013-02-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102938837B (en) | A kind of tone mapping method keeping full sub-model based on edge | |
US10152781B2 (en) | Method for image processing using local statistics convolution | |
Gu et al. | Local edge-preserving multiscale decomposition for high dynamic range image tone mapping | |
CN105850114B (en) | The method of inverse tone mapping (ITM) for image | |
TWI704524B (en) | Method and device for image polishing | |
JP2021077420A (en) | Filtering method, context recognizing type tone mapping operator, and system | |
JP6803378B2 (en) | Reverse tone mapping method and equipment | |
Lee et al. | A space-variant luminance map based color image enhancement | |
CN107408296A (en) | Real-time noise for high dynamic range images eliminates and the method and system of image enhaucament | |
Talebi et al. | Fast multilayer Laplacian enhancement | |
JP2015507241A (en) | Digital makeup | |
CN113822830B (en) | Multi-exposure image fusion method based on depth perception enhancement | |
CN111105359A (en) | Tone mapping method for high dynamic range image | |
Chen et al. | Tone reproduction: A perspective from luminance-driven perceptual grouping | |
Kim et al. | Tone mapping with contrast preservation and lightness correction in high dynamic range imaging | |
CN103020998A (en) | Tone mapping method based on edge-preserving total variation model | |
Krishnamoorthy et al. | Extraction of well-exposed pixels for image fusion with a sub-banding technique for high dynamic range images | |
Thai et al. | Performance evaluation of high dynamic range image tone mapping operators based on separable non-linear multiresolution families | |
Lee et al. | Local tone mapping using sub-band decomposed multi-scale retinex for high dynamic range images | |
CN111815729A (en) | Real-time skin beautifying method, device, equipment and computer storage medium | |
Jia et al. | Efficient and adaptive tone mapping algorithm based on guided image filter | |
CN115937024A (en) | Multi-frame fusion low-illumination image enhancement method based on Retinex theory | |
CN110517210B (en) | Multi-exposure welding area image fusion method based on Haar wavelet gradient reconstruction | |
CN113793247A (en) | Ornament image beautifying method and terminal | |
Goshtasby | High dynamic range reduction via maximization of image information |
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 |