How to quantize the output?

Hello, in gaborconvolve.m code,the output EO consists of real and imaginary components(where real part is the result of convolving with the even symmetric filter, the imaginary part is the result from convolution with the odd symmetric filter). Now, the function must be phase quantized to get output in the form of 0-1 binary matrix...How do i change the code to phase quantize the EO output....

Answer 1

[Walter Roberson]

Supposing you have a routine QUANTIZE, then

output = complex( QUANTIZE(real(E0)), QUANTIZE(imag(E0)) )

[Jyothi Alugolu]

actually the EO consists of complex numbers of size 4 * 6..Now, i want 0-1 binary matrix from real and imaginary components of EO by phase quantization... i am sending you the screenshot regarding output EO..

Now, please tell me how to do this..

[Walter Roberson]

Under what conditions should it produce a binary 0? Under what conditions should it produce a binary 1 ?

[Jyothi Alugolu]

Normalized iris image is transformed into a 0–1 form of binary matrix by convolving 1-D Log-Gabor filter to the normalized image(rectangular strip). Each row in the normalized iris image is considered as a 1-D signal for convolution. The function (here EO) produces real and imaginary components which are phase quantized to get IrisCode in the form of 0–1.

[Jyothi Alugolu]

I think u didn't get me. Actually a rectangular strip has to be transformed into a 0-1 form of binary matrix by convolving 1-D Log-Gabor filter to the normalized image..so, i applied gaborconvolve function to rectangular strip which produces EO which consists of real and imaginary components as i have shown you by screeenshot..Now, the real and imaginary components are phase quantized in the form of 0-1 binary matrix..

[Walter Roberson]

Under what conditions should it produce a binary 0? Under what conditions should it produce a binary 1 ?

For example should it be binary 0 if it is in quadrant 1 or 3, and binary 1 if it is in quadrant 2 or 4?

[Jyothi Alugolu]

I dont know what exactly you are talking..i am implementing a paper..In that paper, it is asked to apply phase qunatasation on complex number to get 0-1 binary matrix..

[Jyothi Alugolu]

in gaborconvolve.m file,we must apply phase quantization to the output to form 0-1 binary matrix

[Jyothi Alugolu]

if amplitude is close to zero then phase data is not useful, so mark off in the noise mask otherwise put it in the template mask..

[Jyothi Alugolu]

i used output = complex( QUANTIZE(real(E0)), QUANTIZE(imag(E0)) ); i got the output..but i am not knowing what it is.. i am sending you the screenshots..please tell me what the output is..

<<

>>

[Walter Roberson]

I have not read the paper, and I am not going to bother to find a free copy of the paper and read through it. You should post examples of inputs and the desired outputs. If there are equations to determine the output from the input phase then you should post them.

Assume that the volunteers are busy people who do not have time to read the various background papers. Post the requirements in clear terms. What you have posted so far is not clear at all.

[Jyothi Alugolu]

ok, i will tell you clearly... i have rectangular iris strip(normalized image) and i want this strip to be transformed into 0-1 binary matrix by convolving 1-D Log-Gabor filter to the normalized image... i used gaborconvolve.m file code and the output EO consists of 4* 6 cell size,where each size consists of 58*212 complex double values(real and imaginary components)..Now, after that the output i.e EO is phase quantized to form 0-1 binary matrix...previously you asked to which quadrant to use..i dont know all that..i just want the rectangular strip to be in0-1 bianry matrix by using 1-D log-gabor filter..

[Walter Roberson]

Result = cellfun( @(M) angle(M) >= 0, EO, 'uniform', 0)

[Jyothi Alugolu]

thank you soo much..