- 學位論文
基於像素空間與亮度值相關性之影像動態範圍擴增技術
LDR to HDR Conversion Based On Spatial And Intensity Correlation
摘要
現今社會的生活品質提升，人們對於影像顯示的要求越來越高，希望將影像呈現接近人眼所看到的真實場景。自然界亮度的動態範圍相當大，但受限於影像格式儲存的關係，影像被壓縮而無法完整呈現真實場景的樣貌。初期的研究方法，對相同場景使用不同曝光度進行固定式拍攝，將多張影像合成為單張高動態範圍影像。雖然合成後高動態範圍影像呈現良好的效果，但是在拍攝過程中可能需要克服目標物的移動與環境變化，否則將造成合成結果不一致。近年來，許多研究希望拍攝過程簡單方便化且得到HDR影像也不亞於用多張影像合成的結果。本篇是屬於這類的研究論文。 本論文提出一套新方法輸入單張低動態範圍影像來產生高動態範圍影像。先將輸入影像轉至HSI色彩空間，對影像的顏色資訊與亮度資訊做分離，並針對影像亮度做處理。接著將亮度的動態範圍由原本8位元延伸至10位元，加入巴特沃斯濾波器形變直方圖後得到新直方圖，將新直方圖設定為目標直方圖，進行配置新影像的處理。在新影像的配置過程中，我們考慮影像像素間的空間距離與像素亮度值差距關係，並使用連通管原理進行像素搬移。接著對影像直方圖計算累積機率函數曲線，且對曲線做左右延伸處理。每回合以固定量做延伸，得到延伸後的新直方圖，再進行新影像配置。最後將新影像結合原始影像彩色資訊，使用反色調對應技術擴展至高動態範圍影像。每回合得到的結果與標準高動態範圍影像進行三種評估分析：HDR-VDP2影像評估、色調對應技術顯示於螢幕進行視覺評估與顯示局部動態範圍分析影像的亮暗區細節並進行分析與討論。
並列摘要
Nowadays, the quality of life has been promoted.People expect the display images to be close to the real scene.In the nature scenes, the dynamic range is broader, but it is limited in the images and cannot be displayed perfectly due to the storage format.In the previous studies, researchers develop the equipment to capture images by using different exposure time at same area and combined them to a high dynamic range image (HDRI).Although the result is good, the dynamic object and environment may cause inconsistent results.In recent years, many researchers want to get good HDR results easily and conveniently.In this thesis we present a new method for generating HDRI.We transform an input image into HSI color space, separating the color and intensity information, and process the intensity channel.Then the dynamic range of intensity is extended from 8 bits to 10 bits and a Butterworth Filter is applied to the histogram which is set as the target histogram to generate a new image allocation.In the new image allocation, we consider image pixel distance and intensity value correlation and use connected-component to set the pixel value.Cumulative Distribution Function Curve of the histogram extends a fixed distance in each round and generates a new histogram which set as a target histogram.Finally,the image is combined with the color information and an inverse tone mapping operator is used to extend the LDR to HDR range.