Presentation is loading. Please wait.

Presentation is loading. Please wait.

Encoding Stereo Images Christopher Li, Idoia Ochoa and Nima Soltani.

Similar presentations


Presentation on theme: "Encoding Stereo Images Christopher Li, Idoia Ochoa and Nima Soltani."— Presentation transcript:

1 Encoding Stereo Images Christopher Li, Idoia Ochoa and Nima Soltani

2 Outline System overview Detailed encoder description Demonstration Results Extensions Conclusions

3 System Overview (Encoder) L DWT Quant Arith Enc DWT Quant Motion Estimation DCT Re- order Arith Enc Huff Enc residuals shift vectors use ME

4 Left Image Daubechies-4 wavelet decomposition – 5 levels for luminance, 4 for chrominance Uniform quantization with adaptive levels – Each component meets its own fraction of MSE Arithmetic coding on the quantized residuals – Frequency tables are sent for each arithmetic coder

5 Left Quantization

6 Left Quantization Motion Estimation Enable Signal Heuristically choose differential vs. separate encoding of right image Calculate MSE Y wavelet coeffs Encode differentially Encode separately Yes No

7 Right Image Motion Estimation Block

8 Right Image Residual coding Impose residuals of Cb and Cr to be 0 – Use remaining fraction of MSE for Y component Compute DCT of block – Reshape using zig-zag ordering – Replace remaining zeros in block with end of block character Perform arithmetic coding

9 Right Image Shift vector coding Offline – Find joint statistics of the shift vectors over the training set – Construct Huffman table During run-time, encode shift vectors using this Huffman table

10 Right Image Separately coded Same method as left image – D4 wavelet, with 5 levels for Y, 4 for Cb, Cr – Uniform quantization with variable step – Arithmetic coding with frequencies sent

11 Writing to File Unique quantization values encoded in header bits Arithmetic coders – Encode frequencies, output length of sequence and sequence itself Huffman encoders – Length of sequence and sequence itself – Tables stored offline

12 Decoder Perform all the steps of the encoder in reverse 1.Decode left image using inverse DWT 2.Read motion estimation flag for right image If enabled, decode shift vectors and residuals Else, decode using inverse DWT

13 Demonstration

14 Results ImageBits/pixel 11.7324 20.7566 30.1807 40.7164 50.9304 61.2390 71.8104 ImageBits/pixel 81.7188 90.6453 101.7766 110.9341 120.5879 132.1404 142.3837

15 Block size

16 Extensions Use intra-block coding for right image Explore using different wavelets Implement embedded zero trees in C Explore run-length coding further Apply uniform deadzone quantizers

17 Conclusions Important trade-off between bits allocated to shift data and residual data Arithmetic coding outperforms Huffman Reshaping the DCT blocks allows us to use information, such as its size, to our advantage Uniform quantizer is faster, simpler and has less overhead than Lloyd-max quantizers MEX files reduce runtime significantly!

18 Thank you Questions?


Download ppt "Encoding Stereo Images Christopher Li, Idoia Ochoa and Nima Soltani."

Similar presentations


Ads by Google