以方向性小波轉換為基礎之影像與視訊編碼

Abstract

多重解析度方向性轉換包含頻率領域和空間領域兩方面。在頻率領域的多重解析度方向性轉換中，以小波轉換為基礎的輪廓轉換能針對影像中的紋路方向，提供較佳的配合度，因此用於影像編碼，但其缺點為較高的計算複雜度。另一方面在空間領域的多重解析度方向性轉換中，方向可調整性小波轉換會根據影像的紋路選出適合的轉換方向，故可提供較小波轉換較佳的壓縮效果。但位於平滑區域中的相鄰區塊的最佳方向的選擇上，常會選出不一致的最佳方向，因此會造成大量且冗餘的外加資訊。 在本論文中，我們提出了數個改良演算法，用以改良以小波轉換為基礎的輪廓轉換以及方向可調整性小波轉換的編碼效果。我們提出了三個改良演算法來改善採用以小波轉換為基礎的輪廓轉換的編碼方法，特別是在減少其計算量。第一，在方向性轉換上，我們提出了一組長度較短的二維濾波器。第二，我們提出一個以移動平均為基礎的判斷方法，用來選出適合作方向性轉換的小波次頻帶。第三，我們改良算術編碼所採用的狀態表，用來加強以小波轉換為基礎的輪廓轉換的編碼效果並減少計算複雜度。經由實驗模擬，我們可以得到和原本採用以小波轉換為基礎的輪廓轉換的編碼方法接近的影像品質，但可減少92%以上的計算量。而和採用傳統二維小波轉換的編碼方法相比，我們能提供較佳的主觀視覺品質。 我們也提出了另外三個改良演算法來改善方向可調整性小波轉換的編碼效果。第一，我們提出了一個由小區塊擴張到大區塊的方向一致化演算法，並把各別區域中，鄰近的方塊的選擇方向作一致化。第二，我們把所提出的方向一致化演算法，延伸到次取樣及方向可調整性之小波轉換上。第三，我們針對動態預測殘像提出了改良式的次取樣及方向可調整性之小波轉換。經由實驗模擬，我們所提出的方向一致化演算法可減少大約60%的外加訊息，並且在低位元率可以增加照片壓縮效果0.4dB左右。而所提出的改良式的次取樣及方向可調整性之小波轉換可以增加視訊壓縮效果0.1~0.2 dB左右。

The multiresolution directional transform includes two types, frequency domain and spatial domain. In frequency domain, the wavelet-based contourlet transform (WBCT) is adopted for image coding because it matches well image textures of different orientations. However, its computational complexity is very high. In spatial domain, the direction-adaptive discrete wavelet transform (DA-DWT) provides better compression performance than discrete wavelet transform (DWT) for it selects transform directions to match image local texture. However, it often picks up inconsistent directions for neighboring blocks with similar texture and thus results in large but redundant side information. In this dissertation, we propose some enhanced algorithms for improving the coding efficiency of WBCT and DA-DWT. we propose three algorithms to enhance the WBCT coding scheme, in particular, on reducing its computational complexity. First, we propose short-length 2-D filters for directional transform. Second, the directional transform is applied to only a few selected subbands and the selection is done by a mean-shift-based decision procedure. Third, we fine-tune the context tables used by the arithmetic coder in WBCT coding to improve coding efficiency and to reduce computation. Simulations show that, at comparable coded image quality, the proposed scheme saves over 92% computing time of the original WBCT scheme. Comparing to the conventional 2-D wavelet coding schemes, it produces clearly better subjective image quality. We propose another three algorithmes to improve the compression performance of DA-DWT in image and video coding. First, we propose a bottom-up direction alignment algorithm to align the directions of neighboring blocks in local regions. Second, we extend alignment algorithm to the subsampling and direction-adaptive discrete wavelet transform (SA-DWT). Third, we propose a modified SA-DWT (MSA-DWT) to improve the motion-compensated residual coding. Simulations show that the proposed alignment algorithm reduces about 60% side information and improves the image coding gain up to 0.4 dB at low bit rate. The MSA-DWT scheme can also improve about 0.1 ~ 0.2 dB in video coding gain.