在色調保真下以多種亮度直方圖等化法控制飽和度的彩色影像增強研究

近年來，許多的產業對於影像品質的要求越來越高，其中包括對於影像的色彩、飽和度和對比度的嚴格要求。而隨著影像處理技術的長足進步，目前已有許多影像增強法可改善輸入影像的品質，但是我們發現，部分影像增強的演算法在某些情況下並沒有獲得影像品質的改善，或是雖然改善了亮度但是卻發生色彩失真的問題，導致影像所呈現的視覺感受不佳。　　有鑑於此，實驗室學長提出利用亮度直方圖等化與雙因子重組飽和度的彩色影像增強法。此方法能夠利用一個參數u值的調整，使亮度變化量以位移與比例這兩個因子產生不同的組合來控制飽和度的變化，有效地保留原始色調並讓輸出影像在視覺感受上呈現良好的結果。然而本研究發現，利用不同u值變化的增強結果雖然有多元的變化，但是部分u值還是令色彩發生視覺感受不和諧的結果，而且還會發生與拍攝影像時的真實情境不相符合的狀況。　　因此本文提出一個改良的彩色影像增強方法，它同樣在色調不變的條件下，能將輸入影像自動分類成暗影像、亮影像和高反差影像等類型，搭配適合各類型影像的亮度等化法並建議較佳的u值範圍來控制飽和度，同時保留影像的真實狀況。實驗結果顯示我們成功地改善上述的彩色影像增強法，在保留影像真實面貌的條件下，有效地改善影像的亮度和對比度並提升視覺的舒適度，而在主觀心理評量的結果中，本文所提方法的輸出結果滿足了使用者對品質的要求而受到使用者的喜愛。

關鍵字

直方圖等化；飽和度控制；色調維持；情境失真

並列摘要

In recent years, more and more industries have become increasingly demanding on image quality, including more stringent demand on image color, saturation and contrast. With remarkable advances in image processing techniques, many image enhancement methods can be used to improve the image quality. However, we found that some of them fail to improve image quality in some situations, where a method may have improve the luminance contrast of image, yet it also causes a color distortion problem which produces a poor visual sensation. In view of the problem above, one of our previous lab members proposed a strategy involving luminance histogram equalization (HE) and two-factor saturation control, where a parameter u can be adjusted such that the brightness variation can be converted to different combinations of saturation control in terms of the shifting and the scaling factors. The method can effectively preserve the original hue and produce a good visual sensation. However, we found that this method still has some shortcomings. Although using different values of u can give different enhancement results of an image, some selected values of u not only make clash of colors, but also produce the false sensation on the scene when the image was actually taken. Therefore, in this thesis we proposed an improved color image enhancement method, which is also capable of preserving the original hue. The proposed method can classify an input image into dark, bright, and high-contrast types automatically and suggest a most suitable HE technique as well as an appropriate value range of u to control the saturation of the input image while preserving the original sensation on the scene. The experimental results show that the proposed method successfully improves the previous work. Under the condition of preserving the original scene sensation, the proposed method effectively improves the performance of image brightness and contrast, and promotes the visual comfort. As for the subjective psychological assessment, the output results with the proposed method are enjoyed by the users because their quality demands are satisfied.

並列關鍵字

histogram equalization ； saturation control ； hue preserving ； loss of real scene sensation

參考文獻

[1] 陳映綱，“利用亮度直方圖等化與雙因子重組飽和度的彩色影像增強法”，中原大學通訊碩士學位學程畢業論文，民國104年。

[6] Zuiderveld, K. “Contrast limited adaptive histogram equalization,” Academic Press Professional Inc., San Diego, CA, USA, pp.474-485, 1994.

[7] Land, E. and McCann, J., “Lightness and retinex theory,” Journal of the Optical Society of America, Volume 61, pp.1-11, 1971.

[8] Jobson, D. J., Rahman, Z-u and Woodell, G. A., “Properties and performance of a center/surround retinex,” IEEE Transactions on Image Processing, Volume 6, Issue 3, pp. 451-462, March 1997.

[9] Moore, A., Allman, J and Goodman, R. M., “A real-time neural system for color constancy,” IEEE Transactions on Neural Networks, Volume 2, Issue 2, pp. 237-247, Feb. 1991.

國際替代計量

在色調保真下以多種亮度直方圖等化法控制飽和度的彩色影像增強研究

全文下載

主題瀏覽