8 - IPOL Demo - Non-Local means - Duration 03:38

Loading...

From the course by Duke University

Image and Video Processing: From Mars to Hollywood with a Stop at the Hospital

389 ratings

Duke University

Image and Video Processing: From Mars to Hollywood with a Stop at the Hospital

389 ratings

In this course, you will learn the science behind how digital images and video are made, altered, stored, and used. We will look at the vast world of digital imaging, from how computers and digital cameras form images to how digital special effects are used in Hollywood movies to how the Mars Rover was able to send photographs across millions of miles of space. The course starts by looking at how the human visual system works and then teaches you about the engineering, mathematics, and computer science that makes digital images work. You will learn the basic algorithms used for adjusting images, explore JPEG and MPEG standards for encoding and compressing video images, and go on to learn about image segmentation, noise removal and filtering. Finally, we will end with image processing techniques used in medicine. This course consists of 7 basic modules and 2 bonus (non-graded) modules. There are optional MATLAB exercises; learners will have access to MATLAB Online for the course duration. Each module is independent, so you can follow your interests.

From the lesson

Spatial processing

Some of the most basic tools in image processing, like median filtering and histogram equalization, are still among the most powerful. We will describe these and provide a modern interpretation of these basic tools. Students will then become familiar with simple and still popular approaches. We will also include non-local means, a more modern technique that still uses classical tools.

1 - Introduction to image enhancement - Duration 19:11 - Optional break at 08:3319:11

2 - Demo - Enhancement & Histogram modification - Duration 03:533:53

3 - Histogram equalization - Duration 19:56 - Optional breaks at 04:40 and 11:3019:57

4 - Histogram matching - Duration 08:318:31

5 - Introduction to local neighborhood operations - Duration 06:466:46

6 - Mathematical properties of averaging - Duration 11:0011:00

7 - Non-Local means - Duration 07:277:27

8 - IPOL Demo - Non-Local means - Duration 03:383:38

9 - Median filter - Duration 07:207:20

10 - Demo - Median filter - Duration 01:311:32

11 - Derivatives, Laplacian & Unsharp masking - Duration 14:24 - Optional breaks at 05:21 and 11:3314:24

12 - Demo - Unsharp masking - Duration 03:103:11

13 - Gradients of scalar and vector images - Duration 05:575:57

14 - Concluding remarks - Duration 01:121:12

Meet the Instructors

Guillermo Sapiro
Professor
Electrical and Computer Engineering

Hello and welcome back. Now that we know, the underlying

framework of non-vocal means, let's just demo it in real time.

Non-local means denoising works using the resources in the image processing online

journal. This is a website that I have provided a

link in their webpage for this class, is a journal that has a lot of online

algorithms for image processing. And I strongly suggest that you visit it

if you want to play with a lot of the state of the art in image processing.

So the journal has first a full description of the algorithm. It has also

code that you can download and use in your own computer, but also has an online

demo and that's what we are going to do next.

You can upload your own images, or you can basically use the images already

available on the website, just to illustrate and to help you understand the

algorithm better. So let's just pick this image, for

example. And it's here.

And we're going to run the algorithm. We're going to add some noise to the

image to see how non-local means will de-noise that, and we're going to ask you

to run. So, now it's running, we see this.

This algorithm can be made to run very fast, but now it's running on a server

and it's running on an implementation that is not optimized for that.

But here we have the result. So we see, let's just illustrate the

details. We see here is the noisy image.

Basically, the original image with the other noise, and now we see the de-noised

image. Again, this is the noisy image, this is

the de-noised image, and this is the original.

So we see that there are differences, but certainly the de-noised image is much

better than the noisy image. Now, we can actually see the difference

between the de-noised image and the original image.

Once again, this is the noisy image, the de-noised image, the original image,

and the difference. They just run with a different image.

So let's just select a new input here, and we can pick a different image, for

example, these dice. And we can pick a different level of

noise, and just put a lot of noise to see what happens here, and then let's just

run it. Once again, while this is running,

remember, the goal of non-local means is to look for regions,

say, okay, if I want to de-noise this region, I'm going to look for similar

regions around and average them. The larger the noise, the more the

algorithm will take, because it's goign to try to average more and more patches

to help us reduce the noise, the noise even further.

So here is, again, the noisy image the algorithm has finished.

This is the noisy. This is the de-noised image.

This is the original, and this is the difference.

So we think it has done a very, very good job in de-noising this image.

Noisy? De-noise.

Noisy, de-noise. Original image and the difference.

This image doesn't have a lot of texture. And that's why these type of algorithms

work really, really, well. There, when there is a lot of texture,

the algorithm still works very well, but for every denoise image denoising

algorithm, dealing with texture is a bit more difficult.

So let's just conclude this by showing this once again.

Noisy, de-noise, original, and difference.

And once again, you can just go into the ipo.im website and play with this or any

other algorithm that is available there, and you can upload your own images and

see how it works. Thank you, and see you in the next video.

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.

Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.

© 2018 Coursera Inc. All rights reserved.

Download on the App Store

Get it on Google Play