how to segment image without change to binary?

how to segment image without change to binary?

Answer 1

[Guillaume]

See this page for several segmentation methods easily implemented in matlab.

Note that most (all?) segmentation methods involve binarisation of the image at some point even if the final output is not binary. There's nothing wrong with that.

Answer 2

[Meshooo]

One idea is to use the color based segmentation.

Check the following nice explanation from the matlab home page:

http://www.mathworks.com/help/images/examples/color-based-segmentation-using-k-means-clustering.html

Answer 3

[Image Analyst]

nina, I don't know what that means. To segment your image means that you have an image, usually a logical image gotten by thresholding, that says where you have your objects of interest ("foreground objects", usually called blobs) and where you don't (the "background"). Like Guillaume said, this involves binarization and there's nothing wrong with that. I don't care whether you're segmenting on color, intensity, texture, or anything else, basically you process the image in some way that you can get a gray scale image that you can threshold, and that thresholded image is a binary image (of logical data type) and that is the segmented image. I don't know that you can avoid it. Why do you want to? Can you give an example of how segmentation would not involve binarization? Of course there are lots of image processing operations that don't involve binarization (like getting a mean intensity, or recovering a watermark, or reconstructing a 3D image from slices, or estimating a range/distance, or whatever) but they're not "segmentation".