在影像處理中，自動白平衡是用來進行不同色溫下的顏色修正的功能，使結果看起來像在正常光源下的影像，而傳統的白平衡方法在某些情況下不能正確的修正色偏。 針對灰界理論在執行過後可能會喪失色彩的豐富度以及色彩直方圖延伸演算法會有存在過度強化的問題。 在這篇論文中，我們提出了一種利用影像特徵為基礎的演算法，結合灰色世界演算法與彩色直方圖延伸，改進了兩種算法的缺點並保留各
自的優點。
本演算法分成三個部分:首先先將輸入影像做兩種不同的白平衡演算法，得到兩張不同的影像，接著分別擷取兩張影像的特徵，再藉由這些特徵去計算出權重，最後融合成兼具融合前各影像優點的新影像。
實驗結果顯示，我們的演算法能更精確的修正色偏影像中的色彩，並且在CIELAB色彩空間中的數值的比較上皆勝過其他傳統的自動白平衡演算法。

In image processing, auto white balance(AWB) is a function used for color correction under different color temperatures, so that the result looks like under canonical illumination. The traditional auto white balance methods can not correct color cast in some situations. For the Gray World theory may lose color richness after execution and the over-contrast problem for the Color Histogram Stretching method. In this thesis, an auto white balance algorithm based on image characteristics has been proposed, combining the Gray World algorithm and the Color Histogram Stretching algorithm, improving the shortcomings of the two algorithms and preserving their respective advantages.

The architecture of the proposed algorithm composed of three main parts: Firstly, the input image is processed by two different auto white balance algorithms and gets two derived inputs. Secondly, the characteristics of these two inputs are extracted and then are calculated the weights according to these characteristics. Finally, the result is obtained by weighted fusing of input images.

The experiment results demonstrate that the proposed algorithm can more accurately correct the color cast image and outperform other auto white balance algorithms in the comparison of the visual quality and the numerical comparisons.

[1] Z. W.Liao,AnAutoWhiteBalanceAlgorithmbyRGBMeanMatchingand
Histogram Stretching,"Master'sthesis,NationalChengKungUniversity,2012.
[2] SuWang;YeweiZhang;PengDeng;FuqiangZhou,"Fastautomaticwhitebalancing
methodbycolorhistogramstretching," Image andSignalProcessing(CISP),2011
4th InternationalCongresson , vol.2,no.,pp.979,983,15-17Oct.2011
[3] Ching-ChihWeng,Chen,H.andChiou-ShannFuh,"Anovelautomaticwhitebal-
ance methodfordigitalstillcameras," Circuits andSystems,2005.ISCAS2005.
IEEE InternationalSymposiumon , vol.,no.,pp.3801,3804Vol.4,23-26May2005
[4] Wikipedia[Online] https://zh.wikipedia.org/zh-tw/%E8%89%B2%E6%B8%A9
[5] E. LANDandJ.McCANN,"LightnessandRetinexTheory," Journal oftheOptical
SocietyofAmerica, pp.1-111971.
[6] KobusBarnard,LindsayMartin,BrianFunt,andAdamCoath,"DataforColour
Research," Color ResearchandApplication, Volume27,Issue3,pp.148-152,2002.
[7] BrianFunt,ComputationalVisionLabofSimonFraserUniversity,Available:http:
//www.cs.sfu.ca/~colour/data/index.html
[8] Ze-NianLiandMarkS.Drew,"FundamentalsofMultimedia",Ch4,2004
[9] Renno,J.-P.,Makris,D.,Ellis,T.andJones,G.A.,"ApplicationandEvaluation
of ColourConstancyinVisualSurveillance," Visual SurveillanceandPerformance
EvaluationofTrackingandSurveillance,2005.2ndJointIEEEInternationalWork-
shop on, pp.301-308,Oct.2005
[10] Flycan.com[Online]Available: http://www.flycan.com.tw/board/modules/
newbb/viewtopic.php?topic_id=170
[11] what-when-how[Online]Available:http://what-when-how.com/nikon-d300s/
manipulating-focus-and-color-in-nikon-d300s/
[12] JinlongLin,"AnAutomaticWhiteBalanceMethodBasedonEdgeDetection,"
Consumer Electronics,2006.ISCE'06.2006IEEETenthInternationalSymposium
on, pp.1-4
[13] Jun-yanHuo,Yi-linChang,JingWang,Xiao-xiaWei,"Robustautomaticwhite
balance algorithmusinggraycolorpointsinimages," Consumer Electronics,IEEE
Transactionson , vol.52,no.2,pp.541-546,May2006
[14] Hong-KwaiLam,Au,O.C.,Chih-WahWong,"Automaticwhitebalancingusing
adjacentchannelsadjustmentinRGBdomain," Multimedia andExpo,2004.ICME
'04. 2004IEEEInternationalConferenceon , vol.2,no.,pp.979-982Vol.2,June2004
[15] Lam,E.Y.,"Combininggrayworldandretinextheoryforautomaticwhitebalance
in digitalphotography," Consumer Electronics,2005.(ISCE2005).Proceedingsof
the NinthInternationalSymposiumon, pp.134-139,June2005
50