Computer Vision Resources for a Beginner

In this article, I will share details of all the computer vision resources, books, papers, etc., that I followed for this subject. The material is such that it will not only provide an excellent introduction to the subject but will quickly familiarize a new beginner with the latest research happening in the field.

Online Lectures

One must begin reading the subject by taking either a university course or an online one. As far as the latter one is concerned, I have found the course ‘Introduction to Computer Vision‘ available on Udacity to be the best for a newbie. The good thing is that this course is free :). We can complement the learning from it by following a book.

Books for Computer Vision

As Computer Vision is a new field, there is no single good introductory book for it. Each book has its own strength and weakness. Thus one has to read only selected portions/chapters from a book and leave others. I recommend the following books along with chapters to read for a new beginner to Computer Vision:-

1. Computer Vision: Algorithms and Applications by Richard Szeliski:- This book does not delve into depth but provides a good breadth of knowledge. Actually, it is a kind of literature survey. I will suggest reading chapters 1 to 5 of the book for an introduction. Further, one can read it as the pdf of the book is available for free.
2. Computer Vision: A Modern Approach (1st Edition) by Forsyth & Ponce:- I recommend reading only Chapter 5, ‘Analytical Image Features’ of the book to learn about the coordinate system used in Computer Vision and various intrinsic/extrinsic camera parameters.
3. Computer Vision: A Modern Approach (2nd Edition) by Forsyth & Ponce:- One has to read Chapter 7: Stereopsis and Chapter 8: Structure from Motion of the book.
4. Three-Dimensional Computer Vision: A Geometric Viewpoint by Olivier Faugeras:- This was the first good book for Computer Vision and now seems outdated. However, one can go through Chapter 2 and Chapter 3 of the book to get in-depth knowledge of camera modeling and calibration.
5. Computer Vision for Visual Effects by Richard J. Radke:-Read Chapter 5 to learn about Epipolar Geometry and Stereo Correspondence.

Research Papers for Computer Vision

I consider certain papers in Computer Vision to be a landmark. The nice thing about them is that they are easy to follow once we have attended the initial few lectures of the Udacity course.

1. P. Perona and J. Malik, “Scale-space and edge detection using anisotropic diffusion“:-  This paper gave the famous Perona and Malik diffusion equation which were non-linear as opposed to linear or Gaussian filtering in vogue at that time. It is considered a pioneer work and has close to 13000 citations :).
2. Lowe, D. (2004), “Distinctive image features from scale-invariant keypoints,” Intl. J. of Computer Vision:- This paper is famous for a descriptor which is scale and rotation invariant. This paper has close to 46000 citations.
3. Joachim Weickert, “Efficient and Reliable Schemes for Nonlinear Diffusion Filtering,” IEEE Transactions on Image Processing, 1998:- Though Perona and Malik gave a non-linear diffusion equation, using that to construct scale-space was an issue. This paper provides instruments to do that. Many future descriptors for key points are based on it.
4. Herbert Bay, “Speeded-Up Robust Features (SURF)“, Computer Vision and Image Understanding, 2008:- This descriptor is reported to be faster than SIFT descriptor. It has close to 11000 citations.
5. Pablo F. Alcantarilla, “KAZE Features“: – A descriptor based on non-linear scale space as opposed to past descriptors based on linear scale space construction to be scale-invariant.

List of articles

The list of articles under computer vision written by me is as follows:-

That was all I had to say about Computer Vision resources for a beginner. Happy learning!