Introduction 720480 pixel color image 24 bitspixelRGB 30
Introduction § 為何要壓縮? § 720*480 pixel color image * 24 bits/pixel(RGB) * 30 images/second = 249 M bps(bits per second) § 超長片(800 MB),只能存 25秒的短片 § 魔戒,片長 3小時,要336 GB的硬碟 § 宿網 100 Mb/s,要抓 7. 5小時 § ADSL 1. 5 Mb/s,要抓 500小時
Lossy Coding of Still Image (JPEG) 8× 8 DCT Q Entropy encoder Compressed image data Source image 8× 8 IDCT Reconstructed image IQ Entropy decoder Compressed image data
Transform Coding X 0 68 75 60 68 53 81 65 45 34 60 61 57 67 50 X 1 169 188 150 171 131 203 162 112 84 150 151 142 168 125
Transform Coding § 目標: 1. 減少原始訊號的相關性 2. 集中大部分訊號能量 § Transform本身並沒有任何壓縮的效果 § Transform的目的是為了作Quantization § 常見的transform方式:linear transform, fourier transform,cosine transform, wavelet transform,KLT……。
Transform Coding § A sample of discrete cosine transform
Quantization
Zig-zag scan
Video Sequence Motion vector Frame i+1 每張frame獨立壓縮 : Quality最好 壓縮效率最差 有效率的壓縮法:參考前一張frame壓縮 能獨立壓縮的frame稱為Intra frame (I-frame) 參考前一張壓縮的frame稱為 P-frame
Motion Estimation § 如何找出motion vector? Finding minimum SAD (Sum of absolute difference)
Motion Estimation
Video Encoder P-frame I-frame Frame input - DCT residual Q VLC Motion Compensation MV Motion Estmation MV Previous frame IDCT Frame memory Clipping Bitstream
Video Decoder I-frame P-frame Coded Bitstream Reconstructed frame IDCT VLC Motion Compensation I P P P P P |←Group of Pictures→| |← Movie Previous Frame memory … P P →|
- Slides: 17