基於內容顯著性之紋理壓縮

Abstract

由於計算能力的受限，有效率且高品質的紋理壓縮技術是非常重要的,在諸如手機的嵌入式系統上尤為明顯。為了能即時的存取隨機像素，降低抽樣率和S3TC技術是最常見的方法，原因在於他們在每個局部區域中都會有固定的壓縮比例。 然而，此方法是均勻的摒棄像素資訊，不會因內容的不同而有所調整。因此我們將展現一個基於內容顯著性的形變來降低紋理的解析度，也就是較均質的區域會被壓縮到較多，以保持更多其他區域中的結構資訊。此外我們將把3D模型上的紋理座標做重新的計算，如此一來在render時將不會有額外的計算成本。 我們的紋理壓縮技術可以和其他方法(例如S3TC)一起實施，因為這些技術在實施過程中是獨立的。藉由一起實施這些技術我們可以達到更高的壓縮比率且遺失較少的資訊以維持視覺上的品質。

Efficient and high quality texture compression is important because computational powers are always limited, especially for embedded systems such as mobile phones. To access random texels instantly, down sampling and S3TC are the most commonly used methods due to the fixed compression ratio at every local region. However, the methods are content oblivious, which uniformly discard texel information. Therefore, we present content aware warp to reduce a texture resolution, where homogeneous regions are squeezed more to retain more structural information. We then relocate texture coordinates of the 3D model as the way of relocating texels, such that rendering the model with our compressed texture requires no additional de-compression cost. Our texture compression technique can cooperate with existing methods such as S3TC since the compression strategies are independent. By reducing texture resolution in a content aware manner, followed by texture compression using S3TC, as the experiments show, we achieve higher compression ratios without losing too much visual quality.