1 / 12

Regression-Based Prediction for Artifacts in JPEG-Compressed Images

Regression-Based Prediction for Artifacts in JPEG-Compressed Images. Park,Jungjin. Introduction. To achieve high compression ratio in JPEG and MPEG, the original image or video may be distorted by blocking and ringing artifact. Goal. Reduction artifacts Reduction time to process

malory
Download Presentation

Regression-Based Prediction for Artifacts in JPEG-Compressed Images

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Regression-Based Prediction for Artifacts in JPEG-Compressed Images Park,Jungjin

  2. Introduction • To achieve high compression ratio in JPEG and MPEG, the original image or video may be distorted by blocking and ringing artifact.

  3. Goal • Reduction artifacts • Reduction time to process • Reduction computational complexity • Simple algorithm

  4. Block Diagram 8x8 DCT Low-Pass Filtering Regression- Based Predicting DCT IDCT 3x3 Gaussian filter could reduce the blocking artifacts. Results in undesirable blurring of filtered image.

  5. Classifier •To classify textures, details, and edges of each DCT block •Calculate the local variable from the DCT coefficients Each DCT coefficient of the DCT block is classified into two distinct classes, CLASS1 CLASS2 Class1 Class2 u,v=1,..8

  6. Threshold

  7. Regression Model with slope Class 2 Without classifier Gauss-Newton method can find the lest square fit estimate of coefficients using linearization Class 1

  8. Regression Model with slope Class1 has a larger slope than Class2 case The slope in the predictors can control the effect of the low-pass filtering 1)Slope>1 image becomes smoother than low pass filtering image 2)Slope<1 image can alleviate the undesirable blurring

  9. Result (reducing blocking) Original test image Recovered image

  10. Result (reducing ringing)

  11. Result (reducing ringing)

  12. Conclusion • Regression based 4.1400 • POCS based 17.1880

More Related