距离变换 matlab实现 opencv实现

最新推荐文章于 2024-04-17 22:10:35 发布

todayq

最新推荐文章于 2024-04-17 22:10:35 发布

阅读量5.3k

点赞数

分类专栏： code 数字图像处理

code 同时被 2 个专栏收录

91 篇文章

订阅专栏

数字图像处理

72 篇文章

订阅专栏

距离变换将一幅二值图像（特征点为0，非特征点为1）变换为一幅灰度图像，灰度图像每一点的像素值为二值图像上非零像素到最近的0像素的距离。

Syntax

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

Description

D = bwdist(BW) computesthe Euclidean distance transform of the binary imageBW.For each pixel in BW, the distance transform assignsa number that is the distance between that pixel and the nearest nonzeropixel ofBW. bwdist uses theEuclidean distance metric by default. BW can haveany dimension.D is the same size as BW.

[D,IDX] = bwdist(BW) alsocomputes 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 samesize as BW and D. Each elementof IDX contains the linear index of the nearestnonzero pixel ofBW.

[D,IDX] = bwdist(BW,method) computesthe distance transform, wheremethod specifiesan alternate distance metric. method cantake any of the following values. Themethod stringcan be abbreviated.

Method	Description
`'chessboard'`	In2-D, the chessboard distance between (x₁,y₁) and (x₂,y₂)is max(│x₁ –x₂│,│y₁ – y₂│).
`'cityblock'`	In2-D, the cityblock distance between (x₁,y₁) and (x₂,y₂)is │x₁ –x₂│+ │y₁ – y₂│.
`'euclidean'`	In2-D, the Euclidean distance between (x₁,y₁) and (x₂,y₂)is This is the default method.
`'quasi-euclidean'`	In2-D, the quasi-Euclidean distance between (x₁,y₁)and (x₂,y₂) is

Class Support

BW can be numeric or logical, and it mustbe nonsparse. D is a single matrix with the samesize asBW. The class of IDX dependson the number of elements in the input image, and is determined usingthe following table.

Class	Range
`'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 representthe position of the nonzero elements using linear matrix indexing.If a pixel contains a 7, its closest nonzero neighbor is at linearposition 7.

Compare the 2-D distance transforms for each of the supporteddistance methods. In the figure, note how the quasi-Euclidean distancetransform best approximates the circular shape achieved by the Euclideandistance 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-Dimage 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')