Alternative Tool to Matlab... Octave in Image Processing

Octave is a free open source program offers services relatively similar to Matlab.

In order to be able to use Octave in Image processing application, you will need to install image package, which can be done easily by typing the following commands in the command window:

Pkg list pkg install -forge general control signal image

·         Each time you want to use image library, you have to load the image package, by typing the following command:

Pkg load image            % only in Octave, in Matlab it is already automatic added

• Some important commands:

Command function	Command syntax
Clear command window	clc
Load image	imread('H:\Octave-4.4.0\samples\myImage.png')
Show an image	imshow(im)
To get the size of a matrix	size(im) > disp(size(im))
To plot some matrices	>> plot(imgreen(210,:))
To display the data type of the matrix	disp(class(im))
To substract 2 images without losing pixels values which will be rounded to 0 because they are negative	Imabsdiff(image1,image2)
To generate a Gausian noise with Standard Deviation between 1 and -1, And you can specify the number of samples To generate a Gausian noise to apply it on an image, we can generate a noise matrix with the same size of our image If we multiply this noise matrix with a scalar then the standard deviation will be changed	Number= randn() // will generate a random number between -1 and 1 >> rand_n=randn() rand_n =  0.083304 >> rand_Vec=randn([1 10]) //will generate a vector of 10 elements between 1 and -1 rand_Vec =    0.242985  -0.150862  -0.856364  -1.788098   0.102084   1.030511   0.267734  -0.102228   0.603328  -0.020853 Noise=randn([1 100]) //generate a vector from 1 to 100 noise=randn(size(peng));
To show how samples are distributed between values, we can use hist where first input is the matrix and second input is the x-axis Hist returns 2 values: N: number of elements in each sample point X: the sample points (x-axis) Then using plot(x,n), we can see how the Gausian distribution is distributed	Noise=randn([1 100]); Hist(noise,[-3 -2 -1 0 1 2 3]); >>  [n,x]=hist(noise,[-3 -2 -1 0 1 2 3]) n =     1   11   23   36   23    5    1 x =   -3  -2  -1   0   1   2   3
To generate number of samples between 2 values, use linspace:	Linspace(-3,3,100) // collect samples between -3 and 3 with 100 sample >> noise=randn([1 1000]); >> [x,n]=hist(noise,linspace(-3,3,21)); >> plot(n,x)
Create a Gaussian filter	h=fspecial('gaussian', kernel_size,sigma);
Apply the filter on the image	out_img=imfilter(img,h);

·