高動態範圍技術於增強內視鏡影像之應用

The Application of High Dynamic Range Technique for Endoscopy Image

潘昭延(Zhao-Yan Pan)

指導教授 ： 蘇振隆

中原大學/工學院/生物醫學工程研究所/碩士(2015年)

https://doi.org/10.6840/cycu201500939

摘要

內視鏡是在檢查消化道疾病時常用的工具，在檢查時必須將光源導入體內，由於是單一光源，離光源較遠區域得到的影像會有漸暗情況，會使醫生在做診斷時耗費不必要的時間更降低效率，本研究主要以影像處理方法、高動態範圍與色調映射為基礎建構一套系統來增加較暗區域，且整體影像並不會有明顯的色調改變，而造成判斷上的誤差。
本研究首先將影像轉至HSV色彩空間後，藉由閥值區的圈選得到較暗區並取得影像資訊，接著以不同組的Gamma值產生出不同的亮度影像，將這些不同的亮度影像結合成高動態範圍影像，經色調映射後增強局區域的亮度。在分析與評估的部分是以不同組的Gamma值比較出最好的亮度區域百分比與系統處理時間，並進而與使用全域運算子、區域運算子、梯度運算子之結果做比較。
本研究使用100張內視鏡影像，由結果可知: 運用不同Gamma值組合之第一組至第四組的提升亮度百分比及處理時間分別為: 21.59%、29.16%、35.28%、39.36% 及16.68秒、19.45秒、19.24秒、27.68秒。比較提昇亮度百分比及處理時間後，第三組Gamma值能夠有效好的結果。與其他色調運算子比較後，本系統所提供的系統效能是優於其他運算子。
本研究所產生的內視鏡影像對於較暗區域的亮度有明顯的提升，且在色調上不會有明顯的改變，在處理時間上無法做即時的顯示，不過大大減少了使用高動態範圍與色調映射所需的時間。另外，影像細節也有明顯的提升，讓醫生在進行檢查時能夠更有效地確認病灶減少診斷所花費的時間。

關鍵字

FICE內視鏡影像 ； Gamma轉換 ； 高動態範圍 ； 色調映射

並列摘要

The endoscopy is always used in diagnosing gastrointestinal disease.
During the detection, light has to induce into body by endoscopy. Because of single light source, there is a part of darker area in the endoscopy image. It will force doctor to waste a lot of time in finding disease. In this system, we combine the gamma correction, high dynamic range and tone-mapping to increase the lightness of darker area. It can also reduce error of judgments due to the change of hue.
The main point of high dynamic range is to generate several images with different brightness or different exposure values to map the tone to low dynamic range image. First of all, the RGB color space was changed to HSV color space. The darker areas would acquire by setting two thresholds. Secondly, different brightness images are combined to high dynamic range image. After the tone-mapping, the lightness of darker area was increased. In order to find the optimal result, four groups with different combination of gamma numbers were used. Moreover, the performance (increased lightness percentage and process time) for other three tone-mapping methods was calculated to compare with the best group in our system.
In this study, 100 endoscopy images were used to test and evaluate this developed system. The increased lightness percentage and process time of the group 1, 2, 3, and 4 are 21.59%, 29.16%, 35.28%, 39.36% and 16.68s, 19.45s, 19.24s, 27.68s, respectively. We could figure out the performance of group 3 was better than other groups. Moreover, after compared with other tone-mapping methods, our system could provide better performance.
Our system could obviously enhance the lightness in the darker areas of endoscopy images. It could reduce the variation of hue and the process time when using high dynamic range and tone-mapping methods. As to increasing the visual ability of contour, our system could provide significant difference which enabled doctor to reduce the time of diagnostic.

並列關鍵字

FICE Endoscopy image ； Gamma correction ； High-dynamic-range ； Tone Mapping

參考文獻

[29]孫善群, ＂利用頻譜分析胃腸道系統影像之輔助診斷系統. ＂ 中原大學生物醫學工程學系碩士論文, 2013.

[30]張孝慈, ＂上消化道診斷之超解析度放大技術之開發. ＂ 中原大學生物醫學工程學系碩士論文, 2015.

[2] Aydin, T. O., Stefanoski, N., Croci, S., Gross, M. and Smolic, A. "Temporally coherent local tone mapping of HDR video." ACM Transactions on Graphics (TOG) 33.6:, pp.196, 2014.

[4] Debevec, Paul E. and Jitendra Malik. "Recovering high dynamic range radiance maps from photographs." ACM SIGGRAPH 2008 classes. ACM, 2008.

[5] Drago, F., Myszkowski, K., Annen, T. and Chiba, N. "Adaptive logarithmic mapping for displaying high contrast scenes." Computer Graphics Forum. Vol. 22. No. 3. Blackwell Publishing, Inc, 2003.