Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

What is a maximum filter response in MR8 filter bank?

So I am implementing The Maximum Response (MR) Filter Bank - MR8. I am using this as a reference: http://www.robots.ox.ac.uk/~vgg/research/texclass/filters.html

And so far I have written the 38 filters, but have no idea how to compute the maximum filter response from the 6 rotations of each edge and bar filter variations.

  1. What is a filter response? Is it the result from the filter application or something else?

  2. What is a maximum response?

  3. How to compute the maximum responses for the edge and bar filters?

like image 910
Rumen Avatar asked Jan 22 '14 15:01

Rumen


People also ask

What is perfect reconstruction filter bank?

A Perfect Reconstruction (PR) filter bank is any filter bank whose reconstruction is the original signal, possibly delayed, and possibly scaled by a constant [287]. In this context, critical sampling (also called ``maximal downsampling'') means that the downsampling factor is the same as the number of filter channels.

How does a filter bank work?

How does a Filter Bank work? A filter bank separates or splits the input signal into multiple components. The process of separating the input signals into multiple components is known as analysis. The output of the analysis is referred to as a sub-band of the original signal.


1 Answers

  1. Suppose you have a filter f and an image I. Filtering I with f is a straightforward convolution of I and f. Let's say the filtered image is I_f. The response of the filter f at a pixel (x,y) of I is then I_f(x,y).

  2. Now, you have a number of oriented filters f1, f2, f3, ... Consequently, you will have filtered images I_f1, I_f2, I_f3, ... From these you construct a maximum response image I_f_max as:

    I_f_max(x,y) = max (I_f1(x,y), I_f2(x,y), I_f3(x,y), ...)

    Therefore, from the application of 6 oriented filters, you have constructed a single response image, which takes into account only the maximum filter response at each pixel and discards the rest.

  3. 1 and 2 should answer this.

like image 166
Papouh Avatar answered Sep 30 '22 18:09

Papouh