find gradient using sobel mask operator in matlab

Question

I am using following code to mask image using sobel mask .

for i=1:size(C,1)-2
    for j=1:size(C,2)-2
        %Sobel mask for x-direction:
        Gx=((2*C(i+2,j+1)+C(i+2,j)+C(i+2,j+2))-(2*C(i,j+1)+C(i,j)+C(i,j+2)));
        %Sobel mask for y-direction:
        Gy=((2*C(i+1,j+2)+C(i,j+2)+C(i+2,j+2))-(2*C(i+1,j)+C(i,j)+C(i+2,j)));

        %The gradient of the image
        %B(i,j)=abs(Gx)+abs(Gy);
        B(i,j)=sqrt(Gx.^2+Gy.^2);
        direction = atan(Gy./Gx)

    end
end

My question is that sometimes gradient direction value is giving as "NaN", how to avoid it?

Second, how to quantize gradient direction into eight zone and find feature vector for the image? Please someone help me.

Why don't you use BW = edge(I,'Sobel') ?? It is so simple. Second, you want to implement HOG algorithm? — Konstantinos Monachopoulos, Feb 25 '17 at 23:07
using sobel technique i will get direction. using this I finally want feature vector then how can I get — Diwakar Kumar, Feb 26 '17 at 07:20

score 0 · Accepted Answer · answered Feb 26 '17 at 15:29

0

Here is the implementation :

clc;clear;close;

% Read the Image
image=imread('girl.png');
f=rgb2gray(image);

%   Initializations
Gx=zeros(size(f,1),size(f,2));
Gy=zeros(size(f,1),size(f,2));
Theta=zeros(size(f,1),size(f,2));
Edges=zeros(size(f,1),size(f,2));

% Sobel Filtering
for x=2:size(f,1)-2
    for y=2:size(f,2)-2

    Gy(x,y)= ( f(x-1,y+1) + 2*f(x,y+1) + f(x+1,y+1) )...
        -( f(x-1,y-1) + 2*f(x,y-1) + f(x+1,y-1) );

    Gx(x,y)= ( f(x+1,y-1) + 2*f(x+1,y) + f(x+1,y+1) )...
        -( f(x-1,y-1) + 2*f(x-1,y) + f(x-1,y+1) );   

    Theta(x,y)= atan(Gx(x,y)/Gy(x,y));  % Direction

    Edges(x,y)=sqrt( Gx(x,y)^2 + Gy(x,y)^2);

    end
end

Results :

answered Feb 26 '17 at 15:29

Konstantinos Monachopoulos

1,045
2
14
29

This is OK thank you for this.But I want feature vector using direction and magnitude so that I can use that for supervised learning.Please tell me how to get feature vector? – Diwakar Kumar Feb 26 '17 at 23:54
What you need is a HOG descriptor, you have to create a histogram of orientations for every cell and pass that to a multi-layer neural network for training. Read this http://users.utcluj.ro/~raluca/prs/prs_lab_05e.pdf to understand the steps !! – Konstantinos Monachopoulos Feb 27 '17 at 00:28

find gradient using sobel mask operator in matlab

1 Answers1