Watershed algorithm MATLAB Programming code parsing

Close all;

%%

%step 1: Grayscale image, Color * * *

RGB = Imread (' pears.png ');

I = Rgb2gray (RGB);

Figure;subplot (121)

Imshow (I)

%step 2: Using gradients to achieve image segmentation

% using the Sobel operator for edge detection,

Text (732,501, ' Image courtesy of Corel ', ' FontSize ', 7, ' HorizontalAlignment ', ' right ')

hy = fspecial (' Sobel ');

HX = Hy ';

Iy = IMFilter (double (I), HY, ' replicate ');% implements the linear spatial filter function, a kind of image enhancement method using filtering processing. Its theoretical basis is spatial convolution and spatial correlation. The aim is to improve image quality, including the removal of high frequency noise and interference, image edge enhancement, linear enhancement and de-blurring.

Ix = IMFilter (double (I), HX, ' replicate ');

Gradmag = sqrt (ix.^2 + iy.^2),% modulo

Subplot (122), Imshow (gradmag,[]), title (' Gradmag ')

% direct with watershed

%l=watershed (GRADMAG);

%lrgb=label2rgb (L);

%figure;imshow (LRGB),

%title (' Lrgb ')

%no. Without additional preprocessing, such as the following markup calculations, the direct result of using watershed transformations is often "over-segmented." ”

% below is the mark foreground and background object

The various programs can be applied here to find the foreground marker, which must be connected to the blob of pixels within each foreground object. In this example, you will use morphological techniques called "Open from Reconstruction" and "closed from reconstruction" to "clean". These actions create a flat maximum value that can be used to imregionalmax within each object.

%step 3: Morphological open operation

SE = Strel (' disk ', 20);% Circular structure element

Io = Imopen (I, se);% morphological open operation

Figure;subplot (121)

Imshow (IO), title (' io ')% shows the graph after execution

%step 4: Corrosion and reconstruction

Ie = Imerode (I, se),% corrosion of image

IOBR = Imreconstruct (Ie, I);% Rebuilding the image

Subplot (122); Imshow (IOBR),% display reconstructed image

Title (' Iobr ')

%step 5: Morphological close operation

Ioc = Imclose (Io, se);% morphological close operation

Figure;subplot (121)

Imshow (IOC),

Title (' Ioc ')

%step 6: Image expansion and negation

IOBRD = imdilate (iobr, SE),% expands the image

IOBRCBR = Imreconstruct (Imcomplement (IOBRD), Imcomplement (IOBR));

IOBRCBR = Imcomplement (IOBRCBR),% image negation

Subplot (122); Imshow (IOBRCBR),

Title (' Iobrcbr ')

%%step 7: Get local maximum value

FGM = Imregionalmax (IOBRCBR);% obtains local maximum value

Figure;imshow (FGM),

Title (' FGM ')

%step 8: Display the maximum area on the original

I2 = I;

I2 (FGM) = 255;% The pixel value at local maximum value is set to 255

Figure;imshow (I2),

Title (' FGM superimposed on original image ')% displays the maximum area on the original

Se2 = Strel (Ones (5,5));% build Element

FGM2 = Imclose (FGM, SE2);% off operation

FGM3 = Imerode (fgm2, se2);% corrosion

FGM4 = Bwareaopen (FGM3, 20);% open operation

%step 9: Show the modified maximum area

I3 = I;

I3 (FGM4) = 255;% foreground set to 255

Figure;subplot (121),

Imshow (I3)% shows the modified maximum area

Title (' Fgm4 superimposed on original image ')

% now marks the background, after cleaning the image, IOBRCBR, dark pixels belong to the background, so you can manipulate from a threshold.

%step 10: Converting to a two-value image

BW = IM2BW (IOBRCBR, Graythresh (IOBRCBR));

Subplot (122); Imshow (BW),

Title (' BW ')

The background pixel is black, but ideally, we don't want the background marker to be too close to the edge of our target object. We subdivide by ' skeletal ', divide the distance of the two value image, and then look for the boundary of the watershed.

%step 11:

D = Bwdist (BW);% calculated distance

DL = watershed (D);% Watershed Transformation

BGM = DL = = 0;% to find the split boundary

Figure Imshow (BGM),% shows the split border

Title (' Watershed Ridge Lines (BGM) ')

Gradmag2 = Imimposemin (Gradmag, BGM | fgm4);

L = watershed (gradmag2);% Watershed Transformation

I4 = I;

I4 (imdilate (L = = 0, ones (3, 3)) | bgm | fgm4) = 255;% foreground and boundary disposition 255

Figure Subplot (121)

Imshow (I4)% highlights foreground and boundary

Title (' Markers and object boundaries ')

Lrgb = Label2rgb (L, ' Jet ', ' w ', ' shuffle '),% converted to pseudo-color * * * *

Subplot (122); Imshow (LRGB)% display pseudo-color * * * * *

Title (' Colored Watershed label Matrix ')

Figure Imshow (I),

On

Himage = Imshow (LRGB),% showing pseudo-color image on original

Set (Himage, ' alphadata ', 0.3);

Title (' Lrgb superimposed transparently on original image ')

This article is from the "Gz7seven" blog, make sure to keep this source http://10734047.blog.51cto.com/10724047/1701827

Watershed algorithm MATLAB Programming code parsing