Hello all, am a beginner to image processing,

 Can you just tell me what are the applications of convolution and where to use it in image processing ? for ex:used for sharpen, blur, edge detction etc. pls help me out.
 The next one is using the convolution matrix or kernel. The image can be supervised flexibly based on the kernel matrix. i usually find kernel matrix pre-defined or entered manually. Is there any way to choose kernel matrix as per our requirements?

Answer 1

[Matz Johansson Bergström]

Convolution, both linear and circular, is used in many areas of optics, image processing, signal processing and beyond.

My master's thesis in computer science was about speeding up convolution using FFT. In my work I wanted to simulate the way a patient with cataracts sees the world. Basically I blurred high dynamic range photos using a large modified gaussian kernel. If you are interested in the details, feel free to read my thesis "Study of Con­vo­lu­tion Algo­rithms using CPU and Graph­ics Hardware" http://matzjb.se/project-reports/.

As you say, convolution can be used for blurring, deblurring, sharpening, edge detection but also noise reduction, see CLEAN. Convolution can also be used in template matching link to wikipedia article.

So convolution is used in many areas in optics and image processing, but it can also be used in higher dimensions in fluid dynamic simulations [Computational Fluid Dynamics by Frederic Magoules chapter 3.6.2].

I hope this shed some light on the subject.

[ankith reddy]

Thanks for your answer sir. one more question sir, is there any possibility to use gabor wavelets along with the convolution?. In case if we use it for what purpose.

[Matz Johansson Bergström]

That I do not know, unfortunately.