conv2

Parameters

hrow

a real or complex vector.

hcol

a real or complex vector.

A

a real or complex 2-D array.

B

a real or complex 2-D array.

shape

an optional character string with possible values:

• "full", conv2 computes the full two-dimensional convolution. It is the default value.
• "same", conv2 computes the central part of the convolution of the same size as A.
• "valid", conv2 computes the convolution parts without the zero-padding of A.

C

a real or complex 2-D array.

Description

conv2 uses a straightforward formal implementation of the two-dimensional convolution equation in spatial form.

C=conv2(A,B [,shape]) computes the two-dimensional convolution of the arrays A and B:

• With shape=="full" the dimensions of the resultC are given by size(A)+size(B)+1. The indices of the center element of B are defined as floor((size(B)+1)/2).
• With shape=="same" the dimensions of the resultC are given by size(A). The indices of the center element of B are defined as floor((size(B)+1)/2).
• With shape=="valid" the dimensions of the result C are given by size(A)-size(B)+1) if and(size(A)-size(B))>=0 else C is empty . The indices of the center element of B are defined as [1 1] .

The separable form C=conv2(hrow,hcol,B [,shape])is equivalent to C=conv2(hrow(:)*hcol(:).',B [,shape])

Note that convol2d can be more efficient for large arrays.

Examples

s = [1 2 1; 0 0 0; -1 -2 -1];//Sobel horizontal edge kernel
A = zeros(10,10);A(3:7,3:7) = 1;
conv2(s,A);

//separable form
u=[1;0;-1];v=[1 2 1];// u*v=s
conv2(u,v,A)

See also

• convol2d — discrete 2-D convolution, using fft.
• conv — discrete 1-D convolution.

History