Accelerating the pace of engineering and science

# Documentation Center

• Trial Software
• Product Updates

# bwdist

Distance transform of binary image

## Syntax

D = bwdist(BW)
[D,IDX] = bwdist(BW)
[D,IDX] = bwdist(BW,method)

## Description

D = bwdist(BW) computes the Euclidean distance transform of the binary image BW. For each pixel in BW, the distance transform assigns a number that is the distance between that pixel and the nearest nonzero pixel of BW. bwdist uses the Euclidean distance metric by default. BW can have any dimension. D is the same size as BW.

[D,IDX] = bwdist(BW) also computes the closest-pixel map in the form of an index array, IDX. (The closest-pixel map is also called the feature map, feature transform, or nearest-neighbor transform.) IDX has the same size as BW and D. Each element of IDX contains the linear index of the nearest nonzero pixel of BW.

[D,IDX] = bwdist(BW,method) computes the distance transform, where method specifies an alternate distance metric. method can take any of the following values. The method string can be abbreviated.

Method

Description

'chessboard'

In 2-D, the chessboard distance between (x1,y1) and (x2,y2) is max(│x1x2│,│y1y2│).

'cityblock'

In 2-D, the cityblock distance between (x1,y1) and (x2,y2) is │x1x2│ + │y1y2│.

'euclidean'

In 2-D, the Euclidean distance between (x1,y1) and (x2,y2) is

This is the default method.

'quasi-euclidean'

In 2-D, the quasi-Euclidean distance between (x1,y1) and (x2,y2) is

## Class Support

BW can be numeric or logical, and it must be nonsparse. D is a single matrix with the same size as BW. The class of IDX depends on the number of elements in the input image, and is determined using the following table.

ClassRange
'uint32'numel(BW) <= 232 − 1
'uint64'numel(BW) >= 232

## Examples

Compute the Euclidean distance transform.

```bw = zeros(5,5); bw(2,2) = 1; bw(4,4) = 1
bw =
0     0     0     0     0
0     1     0     0     0
0     0     0     0     0
0     0     0     1     0
0     0     0     0     0

[D,IDX] = bwdist(bw)

D =
1.4142    1.0000    1.4142    2.2361    3.1623
1.0000         0    1.0000    2.0000    2.2361
1.4142    1.0000    1.4142    1.0000    1.4142
2.2361    2.0000    1.0000         0    1.0000
3.1623    2.2361    1.4142    1.0000    1.4142

IDX =
7     7     7     7     7
7     7     7     7    19
7     7     7    19    19
7     7    19    19    19
7    19    19    19    19```

In the nearest-neighbor matrix IDX the values 7 and 19 represent the position of the nonzero elements using linear matrix indexing. If a pixel contains a 7, its closest nonzero neighbor is at linear position 7.

Compare the 2-D distance transforms for each of the supported distance methods. In the figure, note how the quasi-Euclidean distance transform best approximates the circular shape achieved by the Euclidean distance method.

```bw = zeros(200,200); bw(50,50) = 1; bw(50,150) = 1;
bw(150,100) = 1;
D1 = bwdist(bw,'euclidean');
D2 = bwdist(bw,'cityblock');
D3 = bwdist(bw,'chessboard');
D4 = bwdist(bw,'quasi-euclidean');
figure
subplot(2,2,1), subimage(mat2gray(D1)), title('Euclidean')
hold on, imcontour(D1)
subplot(2,2,2), subimage(mat2gray(D2)), title('City block')
hold on, imcontour(D2)
subplot(2,2,3), subimage(mat2gray(D3)), title('Chessboard')
hold on, imcontour(D3)
subplot(2,2,4), subimage(mat2gray(D4)), title('Quasi-Euclidean')
hold on, imcontour(D4)```

Compare isosurface plots for the distance transforms of a 3-D image containing a single nonzero pixel in the center.

```bw = zeros(50,50,50); bw(25,25,25) = 1;
D1 = bwdist(bw);
D2 = bwdist(bw,'cityblock');
D3 = bwdist(bw,'chessboard');
D4 = bwdist(bw,'quasi-euclidean');
figure
subplot(2,2,1), isosurface(D1,15), axis equal, view(3)
camlight, lighting gouraud, title('Euclidean')
subplot(2,2,2), isosurface(D2,15), axis equal, view(3)
camlight, lighting gouraud, title('City block')
subplot(2,2,3), isosurface(D3,15), axis equal, view(3)
camlight, lighting gouraud, title('Chessboard')
subplot(2,2,4), isosurface(D4,15), axis equal, view(3)
camlight, lighting gouraud, title('Quasi-Euclidean')```

## More About

expand all

### Tips

bwdist uses fast algorithms to compute the true Euclidean distance transform, especially in the 2-D case. The other methods are provided primarily for pedagogical reasons. However, the alternative distance transforms are sometimes significantly faster for multidimensional input images, particularly those that have many nonzero elements.

The function bwdist changed in version 6.4 (R2009b). Previous versions of the Image Processing Toolbox used different algorithms for computing the Euclidean distance transform and the associated label matrix. If you need the same results produced by the previous implementation, use the function bwdist_old.

### Algorithms

For Euclidean distance transforms, bwdist uses the fast algorithm described in

[1] Maurer, Calvin, Rensheng Qi, and Vijay Raghavan, "A Linear Time Algorithm for Computing Exact Euclidean Distance Transforms of Binary Images in Arbitrary Dimensions," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, No. 2, February 2003, pp. 265-270.

For cityblock, chessboard, and quasi-Euclidean distance transforms, bwdist uses the two-pass, sequential scanning algorithm described in

[2] Rosenfeld, Azriel and John Pfaltz, "Sequential operations in digital picture processing," Journal of the Association for Computing Machinery, Vol. 13, No. 4, 1966, pp. 471-494.

The different distance measures are achieved by using different sets of weights in the scans, as described in

[3] Paglieroni, David, "Distance Transforms: Properties and Machine Vision Applications," Computer Vision, Graphics, and Image Processing: Graphical Models and Image Processing, Vol. 54, No. 1, January 1992, pp. 57-58.

## See Also

Was this topic helpful?