Image Segmentation

12
Image Image Segmentation Segmentation Chin-Ya Huang Chin-Ya Huang Mon-Ju Wu Mon-Ju Wu ECE 533 Final Project, Fall ECE 533 Final Project, Fall 2006 2006 University of Wisconsin – University of Wisconsin – Madison Madison

description

Image Segmentation. Chin-Ya Huang Mon-Ju Wu ECE 533 Final Project, Fall 2006 University of Wisconsin – Madison. Methodology. Acquire the color information and the edge information separately. Use hue, saturation and intensity to get color information. - PowerPoint PPT Presentation

Transcript of Image Segmentation

Page 1: Image Segmentation

Image Image SegmentationSegmentation

Chin-Ya Huang Chin-Ya Huang Mon-Ju WuMon-Ju Wu

ECE 533 Final Project, Fall 2006ECE 533 Final Project, Fall 2006University of Wisconsin – MadisonUniversity of Wisconsin – Madison

Page 2: Image Segmentation

MethodolMethodologyogy

1)1) Acquire the color Acquire the color information and the edge information and the edge information separately.information separately.

2)2) Use hue, saturation and Use hue, saturation and intensity to get color intensity to get color information.information.

3)3) Use the Matlab “edge” Use the Matlab “edge” command to extract the command to extract the image boundary.image boundary.

4)4) Combine the above result by Combine the above result by getting the union of (2) and getting the union of (2) and (3).(3).

5)5) Final modification.Final modification.

Page 3: Image Segmentation

MethodolMethodologyogy

Original Image

Texture Extraction

Shape Extraction

Intersaction

End

Start

Extracted Edge

Separate image into H. S. I components

Page 4: Image Segmentation

ExampleExample

Target:Target: Extract the Extract the

image of the image of the soccer player soccer player out from the out from the entire image.entire image.

Page 5: Image Segmentation

Step 1 : Using HSI as a Step 1 : Using HSI as a thresholdthreshold

A)A) Cut partial Cut partial images to images to compute the compute the HSI values.HSI values.

B)B) Set the Set the threshold threshold value.value.

C)C) Examine Examine through the through the image to image to acquire the acquire the color color information.information.

Page 6: Image Segmentation

Step 2 : Using Matlab “edge” Step 2 : Using Matlab “edge” commandcommand

A)A) Use Matlab “edge” Use Matlab “edge” command to extract the command to extract the course edges.course edges.

B)B) Use Matlab “imfill” Use Matlab “imfill” command to fill the area command to fill the area whose surrounding course whose surrounding course edges make a closure.edges make a closure.

Page 7: Image Segmentation

edgeedge imfill(edimfill(edge)ge)

Page 8: Image Segmentation

Step 3 : Combine the above Step 3 : Combine the above imagesimages

A)A) Get the Get the union of the union of the images from images from Step 1 and Step 1 and Step 2.Step 2.

B)B) Combine the Combine the color color information information and the and the edge.edge.

Page 9: Image Segmentation

Step 4 : Final modificationStep 4 : Final modification

A)A) Perform Perform dilate on dilate on “edge” make “edge” make sure all the sure all the separate line separate line segments are segments are connected. connected.

B)B) Use “imfill” Use “imfill” command to command to fill the area fill the area inside the inside the boundary.boundary.

Page 10: Image Segmentation

Step 4 : Final modification Step 4 : Final modification (cont.)(cont.)C)C) Intersect the Intersect the

image from image from (B) with the (B) with the image from image from Step 3 to Step 3 to remove the remove the noise.noise.

D)D) Image from Image from (C) is the (C) is the final result.final result.

Page 11: Image Segmentation

Final Final ResultResult

Original Original ImageImage

Page 12: Image Segmentation

Future WorkFuture Work

A)A) Develop a more powerful Develop a more powerful algorithm to perform algorithm to perform interpolation in order to interpolation in order to connect the disconnected connect the disconnected line segment.line segment.

B)B) Find other ways to segment Find other ways to segment image other than using HSI.image other than using HSI.

C)C) Develop a more powerful Develop a more powerful algorithm for noise removal.algorithm for noise removal.