So far, we have heard seven Friday session talks covering a wide range of topics from empirical mode decomposition and diffusion tensor MRI to image watermarking and biometrics. I believe it is fair to say that the diversity of the topics (arising from diverse background of students taking this course) is consistent with the diverse [...]
Archive for October, 2008
Interaction between science and engineering
Posted in wavelets on October 28, 2008 | Leave a Comment »
Exploring the Jungle of PDE-based Image Processing
Posted in ee565 on October 28, 2008 | 1 Comment »
This week’s lectures contain a short journey to a remote area less familiar to engineering students: the jungle of PDE-based image processing. I don’t expect to cover a lot of territories due to the time constraint. But I am hoping the three exemplar PDEs: mean-curvature diffusion, total-variation diffusion and perona-malik diffusion can at least reveal [...]
Challenges in Assessing Subjective Quality of Images
Posted in ee565 on October 27, 2008 | 2 Comments »
In the past week, I have shown you many images generated by different interpolation techniques and ask you to see this artifact or that artifact. And I kept reminding you that MATLAB imshow function or the projection of images onto the screen has caused further distortion to visual inspection. All these are related to a [...]
What is good and bad about exams?
Posted in wavelets on October 27, 2008 | Leave a Comment »
As midterm exam approaches, I start to sense the anxiety among my students. It reminds me how I felt about exams when I was a student. As a student coming from China, I can’t even remeber how many exams I have been through in my school years (in the worst time – say middle school, [...]
Algebraic vs. geometric aspect of wavelets
Posted in wavelets on October 21, 2008 | Leave a Comment »
As we gradually transit from first-generation to second-generation wavelets, the style of thinking varies. In Daubechie’s construction, we heavely used algebraic tricks and motivations such as the polynomial of some special property.Despite the elegance of such construction, it turns out that there are alternative approaches which rely on the language of geometry. Geometry is usually [...]
ICIP2008 Impression
Posted in ee565 on October 21, 2008 | 1 Comment »
Int. Conf. on Image Processing (ICIP) is the flag conference in the technical field of image processing. Since its inception in 1994, ICIP has grown into a major conference attracting thousands of attendees from all over the world. The acceptance rate of this conference is usually 40%, which allows is to have multiple oral and [...]
What is the next killer application?
Posted in ee565 on October 8, 2008 | Leave a Comment »
From our discussion in the class, you have seen how texture synthesis has been widely used in image inpainting and quilting applications. I think it is safe to say that this is one of fastest growing area in computer graphics even though almost all existing techiques are based on the same fundamental idea of nonparametric [...]
Texture Synthesis: Seeing is Believing
Posted in ee565 on October 1, 2008 | Leave a Comment »
The underlying hypothesis of all texture synthesis methods is perceptually similar textures have similar statistics (historically it was the renowned visual neuroscientist B. Julesz who made such conjecture first). In all experiments reported in the paper and in our class, we simply trust our own eyes to judge whether two images (the reference and the [...]
Importance of phase in signal processing and science
Posted in wavelets on October 1, 2008 | Leave a Comment »
As early as in 1981, the importance of phase was recognized by signal processing researchers (Here is the link to the article http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1456290). Note that that was before the infancy of wavelets – therefore the phase in that article solely refers to the use in Fourier transforms. Twenty years after the invention of wavelet transforms, [...]
