Abstract
一種使接收端能獲得高品質的視訊影像,並顯示在各式的顯示器上的方法, 在本論文中被提出。為了使接收端能得到高品質之視訊影像,我們提出一 調適性後處理方法。該方法是利用動態補償及像框平均,來改進影像在時 域上的品質。而在空域上,將每個像點根據其特性,加以分類,再根據每個 像點的特性,施以不同的濾波器,來過濾雜質,以達到影像品質提高的效果 。尤其是屬於邊界之像點,對於人類視覺來說是非常重要,因此,我們使用 多層中質濾波器來過濾在邊界像點之雜訊,並同時保持影像的清析度。為 了顯示影像在有限顏色的顯示器上,我們提出了一個快速非平衡貪婪樹的 方法。此方法是構建在二元樹的建構上。該樹在建構時,一次只會增加一 個節點。節點的增長,是將具有最大誤差縮減的節點,一分為二。為了去除 因顏色數減少,而產生的錯誤輪廓,我們又提出了一像點對應的方法。此方 法是先找出各顏色區的邊界,再對這些邊界區,施以誤差擴散法來消除錯誤 輪廓的效應。而尋找各顏色區的邊界點的方法,則是利用形態學裡的侵蝕 法及相差法來達成。半色調法是將具有灰度值之影像,表現在只具有二個 值的顯視器,如二元螢幕或二元雷射印表機等,的一種方法。使得一個二元 影像,在人眼看來仍具有灰度值之效果。而我們所提之半色調法,乃是利用 離散餘弦轉換來達成。我們先將影像區分成多個子區塊,每個子區塊作離 散餘弦轉換,然後再從二元子區塊裡找出其做離散餘弦轉換後之值,與原始 影像子區塊離散餘弦值最接近。所找出之二元區塊即代表原始影像子區 塊,而顯示在二元影像顯示器上。 The techniques for a receiver to produce good quality video signals and to show the image on the image display are proposed. To produce good quality video signals in the receiver, an adaptive postprocessing algorithm to enhance the quality of a noisy video sequence is proposed. The algorithm employs motion compensated frame averaging to improve picture quality in the temporal domain. A classification algorithm divides subblocks of pixels in the averaged frame into some classes. Spatial algorithm applys different filters to the pixels belonging to different class for preserving the details. Since the edge is very important for human perception, the multi-level median filter is applied to the edge pixels to remove the noise and preserve the edge information. To display the image on the display with limited color palette, a fast unbalanced greedy tree growing is proposed. The algorithm is based on the construction of a binary tree. The tree is designed one node at a time by splitting the node with the largest error reduction. For reducing the false contouring caused by abrupt color changes across region boundaries, a pixel mapping algorithmis proposed. The algorithm extracts the boundary of each color region by morphological erosion filter, and then applys the error diffusion to the boundary to alleviate the false contouring. Halftoning is a technique for rendering the continuous-tone image on the bilevel display. A new algorithm for halftoning based on discrete cosine transform is proposed. The algorithm replaces continuous-tone image subblock with a bilevel pattern which minimize the distortion between the continuous-tone image subblock and bilevel pattern in the frequency domain. From the experimental results, it is clear that the proposed algorithm can produce high quality of halftoned images.