- 學位論文
基於客觀品質最佳化之色調對映技術
Optimal Objective Quality Based Tone Mapping Technique
摘要
近年隨著多媒體的蓬勃發展，各家廠商推出了許多提升觀賞技術的電子產品。影像品質一直是消費者選購的最重要考量因素，解析度、對比度、色彩表現等是提升畫質的重要因素。從最早的一般畫質(Standard Definition，SD)，高畫質(High Definition，HD)，到近年來興起的超高畫質(Ultra High Definition，UHD)，皆代表了每個世紀的新興技術。除了將解析度提高外，增廣色彩色域及對比度，讓電視影像呈現與人眼看到自然的景物接近，也是提升電視畫質方法。高動態範圍（High Dynamic Range，HDR）技術可讓電視達成上述目標，與提高解析度的4K電視比較，HDR是截然不同的技術。在HDR作用下，電視所呈現的色彩更加生動，白色更明亮，黑色更深邃，畫面中的物體也更加清晰，畫面色域也會被擴大，可在冷色和暖色間進行交替。 儘管現今技術已能生產可觀賞HDR的電視，卻依舊面臨兩大問題，一是因為拍攝器材的昂貴價格，以HDR技術拍攝的影片數量極為稀少，二是HDR電視尚未成為主流，現今一般家庭大部分人使用傳統顯示器，因此如何把高動態範圍影像，藉由良好的色調對映技術將動態範圍壓縮至低動態範圍，並顯示出影像對比度與自然度是一大考驗。 本論文以客觀品質量測技術TMQI(Objective Quality Assessment of Tone Mapped Images)為基礎，改善它的客觀品質量測技術，提升與主觀成績的關聯性。首先輸入HDR影像，對HDR影像亮度直方圖進行適當的區域分割，之後改良式 TMQI的成績為依據，搜索LDR上的最佳對應點，以進行HDR各區域的線性對映，達到最佳LDR影像輸出結果，給予觀賞者良好且低成本的視覺體驗。 當今常見的色調對映技術存在著一項缺點，即一種色調對映技術無法完美的處理不同特性之HDR影像，而我們所提出的方法無論輸入的HDR影像特性為何，皆能輸出良好的LDR影像品質。
並列摘要
In recent years, with the rapid development of multimedia, various manufacturers have introduced a number of technologies to enhance the viewing of electronic products. Image quality has been the most important of consideration to consumers. Resolution, contrast and color performance are important factors to enhance the quality of image. From the earliest Standard Definition(SD), High Definition(HD), until recent years, the rise of ultra-high-definition(UHD), all of them representative of the emerging technologies of each century. In addition to the resolution increase, broaden color gamut and contrast. So that TV images showing the human eye to see the natural scenery close, but also a way to enhance TV picture quality. HDR (High Dynamic Range) technology allows the television to achieve the above goals. Compared with improving resolution 4K TV, HDR is a completely different technology. In the HDR effect, TV presented colors more vivid, brighter, and makes picture objects more clear, the screen will be expanded color gamut, can alternate between warm and cool colors. Although the current technology has been able to produce HDR TV, The first reason is because the price of expensive photographic equipment, the number of films shot in HDR technology is extremely rare, and the second reason is that HDR television has not yet become mainstream, Nowadays, most people use traditional displayer. So how to make high dynamic range images, use the well tone mapping technology compress high dynamic range to low dynamic range, And can display the image contrast and naturalness is a major test. In this paper, based on an objective quality measurements technology TMQI (Objective Quality Assessment of Tone Mapped Images), and improve it assessment Accuracy. Improve the relevance of subjective and objective achievement scores. First, Input HDR image, HDR image brightness histogram appropriate regional division. Followed by Modified TMQI results as the basis, Find the best corresponding points on the LDR image, then linear mapping each HDR regions to LDR regions, Until the best LDR image output results found, and giving the viewer a good and cost-visual experience. Nowadays common tone mapping technologies exist a disadvantage, That is, a tone mapping technique cannot be perfect for processing HDR images of different characteristics. Our proposed method no matter what the input of HDR images of characteristic, still can maintain the quality of LDR images.