Compute 2-D discrete convolution of two input matrices
Filtering
visionfilter
The 2-D Convolution block computes the two-dimensional convolution of two input matrices. Assume that matrix A has dimensions (Ma, Na) and matrix B has dimensions (Mb, Nb). When the block calculates the full output size, the equation for the 2-D discrete convolution is
$$C(i,j)={\displaystyle \sum _{m=0}^{(Ma-1)}{\displaystyle \sum _{n=0}^{(Na-1)}A(m,n)*B(i-m,j-n)}}$$
where $$0\le i<Ma+Mb-1$$ and $$0\le j<Na+Nb-1$$.
Port | Input/Output | Supported Data Types | Complex Values Supported |
---|---|---|---|
I1 | Matrix of intensity values or a matrix that represents one plane of the RGB video stream |
| Yes |
I2 | Matrix of intensity values or a matrix that represents one plane of the RGB video stream | Same as I1 port | Yes |
Output | Convolution of the input matrices | Same as I1 port | Yes |
If the data type of the input is floating point, the output of the block has the same data type.
The dimensions of the output are dictated by the Output
size parameter. Assume that the input at port I1 has dimensions
(Ma, Na) and the input at port
I2 has dimensions (Mb, Nb).
If, for the Output size parameter, you choose Full
,
the output is the full two-dimensional convolution with dimensions
(Ma+Mb-1, Na+Nb-1).
If, for the Output size parameter, you choose Same
as input port I1
, the output is the central part of the
convolution with the same dimensions as the input at port I1. If,
for the Output size parameter, you choose Valid
,
the output is only those parts of the convolution that are computed
without the zero-padded edges of any input. This output has dimensions
(Ma-Mb+1, Na-Nb+1).
However, if all(size(I1)<size(I2))
, the block
errors out.
If you select the Output normalized convolution check
box, the block's output is divided by sqrt(sum(dot(I1p,I1p))*sum(dot(I2,I2)))
,
where I1p
is the portion of the I1 matrix that
aligns with the I2 matrix. See Example 2 for
more information.
Note: When you select the Output normalized convolution check box, the block input cannot be fixed point. |
The following diagram shows the data types used in the 2-D Convolution block for fixed-point signals.
You can set the product output, accumulator, and output data types in the block mask as discussed in Parameters.
The output of the multiplier is in the product output data type if at least one of the inputs to the multiplier is real. If both of the inputs to the multiplier are complex, the result of the multiplication is in the accumulator data type. For details on the complex multiplication performed, refer to Multiplication Data Types.
Suppose I1, the first input matrix, has dimensions (4,3) and
I2, the second input matrix, has dimensions (2,2). If, for the Output
size parameter, you choose Full
, the
block uses the following equations to determine the number of rows
and columns of the output matrix:
The resulting matrix is
If, for the Output size parameter, you
choose Same as input port I1
, the output is the
central part of $$Cfull$$ with
the same dimensions as the input at port I1, (4,3). However, since
a 4-by-3 matrix cannot be extracted from the exact center of $$Cfull$$,
the block leaves more rows and columns on the top and left side of
the $$Cfull$$ matrix
and outputs:
If, for the Output size parameter, you
choose Valid
, the block uses the following equations
to determine the number of rows and columns of the output matrix:
In this case, it is always possible to extract the exact center of $$Cfull$$. Therefore, the block outputs
In convolution, the value of an output element is computed as a weighted sum of neighboring elements.
For example, suppose the first input matrix represents an image and is defined as
I1 = [17 24 1 8 15 23 5 7 14 16 4 6 13 20 22 10 12 19 21 3 11 18 25 2 9]
The second input matrix also represents an image and is defined as
I2 = [8 1 6 3 5 7 4 9 2]
The following figure shows how to compute the (1,1) output element (zero-based indexing) using these steps:
Rotate the second input matrix, I2, 180 degrees about its center element.
Slide the center element of I2 so that it lies on top of the (0,0) element of I1.
Multiply each element of the rotated I2 matrix by the element of I1 underneath.
Sum the individual products from step 3.
Hence the (1,1) output element is $$0\cdot 2+0\cdot 9+0\cdot 4+0\cdot 7+17\cdot 5+24\cdot 3+0\cdot 6+23\cdot 1+5\cdot 8=220$$.
Computing the (1,1) Output of Convolution
The normalized convolution of the (1,1) output element is 220/sqrt(sum(dot(I1p,I1p))*sum(dot(I2,I2)))
=
0.3459, where I1p = [0 0 0; 0 17 24; 0 23 5]
.
This parameter controls the size of the output scalar, vector,
or matrix produced as a result of the convolution between the two
inputs. If you choose Full
, the output has dimensions
(Ma+Mb-1, Na+Nb-1).
If you choose Same as input port I1
, the output
has the same dimensions as the input at port I1. If you choose Valid
,
output has dimensions (Ma-Mb+1, Na-Nb+1).
If you select this check box, the block's output is normalized.
Select the rounding mode for fixed-point operations.
Select the Overflow mode for fixed-point operations.
Use this parameter to specify how to designate the product output word and fraction lengths. Refer to Fixed-Point Data Types and Multiplication Data Types for illustrations depicting the use of the product output data type in this block:
When you select Same as first input
,
these characteristics match those of the first input to the block.
When you select Binary point scaling
,
you can enter the word length and the fraction length of the product
output, in bits.
When you select Slope and bias scaling
,
you can enter the word length, in bits, and the slope of the product
output. The bias of all signals in the Computer Vision System Toolbox™ software
is 0.
The Product Output inherits its sign according to the inputs. If either or both input I1 and I2 are signed, the Product Output will be signed. Otherwise, the Product Output is unsigned. The following table shows all cases.
Sign of Input I1 | Sign of Input I2 | Sign of Product Output |
---|---|---|
unsigned | unsigned | unsigned |
unsigned | signed | signed |
signed | unsigned | signed |
signed | signed | signed |
Use this parameter to specify how to designate the accumulator word and fraction lengths. Refer to Fixed-Point Data Types and Multiplication Data Types for illustrations depicting the use of the accumulator data type in this block. The accumulator data type is only used when both inputs to the multiplier are complex:
When you select Same as product output
,
these characteristics match those of the product output.
When you select Same as first input
,
these characteristics match those of the first input to the block.
When you select Binary point scaling
,
you can enter the word length and the fraction length of the accumulator,
in bits.
When you select Slope and bias scaling
,
you can enter the word length, in bits, and the slope of the accumulator.
The bias of all signals in the Computer Vision System Toolbox software
is 0.
Choose how to specify the word length and fraction length of the output of the block:
When you select Same as first input
,
these characteristics match those of the first input to the block.
When you select Binary point scaling
,
you can enter the word length and the fraction length of the output,
in bits.
When you select Slope and bias scaling
,
you can enter the word length, in bits, and the slope of the output.
The bias of all signals in the Computer Vision System Toolbox software
is 0.
Select this parameter to prevent the fixed-point tools from
overriding the data types you specify on the block mask. For more
information, see fxptdlg
,
a reference page on the Fixed-Point Tool in the Simulink^{®} documentation.
Computer Vision System Toolbox software |