Project Three

This Project counts 20% of the total score. Each student should choose only one topic among the following three and implement it by yourself.

Topic One: Denoising

1. Add different types of noise (salt-and-pepper noise, Gaussian noise) to test images and denoise the images with basic filtering methods including smoothing linear filters, median filters, frequency domain filters, etc. Change the noise adding and filtering parameters and compare the effects of different methods.

2. Implement Block Matching and 3D Filtering for image denoising.

3. You should complete the project on your own. Package your code and report in a folder named ‘Pro3_number_name’.

4. Submitting: ftp://public.sjtu.edu.cn

username: xiongyuehan

password: public

Please put your file under directory upload\project_3\topic_1. If that fails, please send it to xiongyuehan@sjtu.edu.cn

deadline: 30th May, 2016

5. Reference

1) test_image3

2) Block Matching and 3D Filtering
Image denoising with block-matching and 3D filtering
Image denoising by sparse 3D transform-domain collaborative filtering

Topic Two: Image Super-Resolution
1. Implement image super-resolution algorithm presented in the paper "Image Super-Resolution as Sparse Representation of Raw Image Patches" by yourself.

2. Your report should at least include the algorithm principles, implementation process, experimental results and the analysis of them.

3. You should complete the project on your own. Package your code and report in a folder named ‘Pro3_number_name’.

4. Submitting: ftp://public.sjtu.edu.cn

username: william-g

password: public

Please put your file under directory upload\project3. If that fails, please send it to william-g@sjtu.edu.cn

deadline: 30th May, 2016

5. Reference

1) test_image4
2)Image Super-Resolution as Sparse Representation of Raw Image Patches

Topic Three: Edge Detection and Segmentation

1. Implement Mean Shift algorithm for edge detection and segmentation on test images. Change the parameters and compare segmentation results, point out the proper range of parameters.

2. Your report should at least include the algorithm principles, implementation process, experimental results and the analysis of them.

3. You should complete the project on your own. Package your Matlab code and report in a folder named ‘Pro3_number_name’.

4. Submitting: ftp://public.sjtu.edu.cn

username: xiongyuehan

password: public

Please put your file under directory upload\project_3\topic_3. If that fails, please send it to xiongyuehan@sjtu.edu.cn

deadline: 30th May, 2016

5. Reference

1) test_image5

2) Mean Shift
Mean Shift: A Robust Approach Toward Feature Space Analysis