Is there any method that automates the segmentation without the human intervention?If so please provide the details of the method.

Currently i am using k-means clustering method for segmentation.But it requires human intervention. I want to use automated method which should provide same results as k-means clustering. Please provide the solution for this.

Answer 1

[Walter Roberson]

http://www.mathworks.com/matlabcentral/fileexchange/36634-an-efficient-technique-for-white-blood-cells-nuclei-automatic-segmentation

http://www.mathworks.com/matlabcentral/fileexchange/53114-automated-lung-segmentation

http://www.mathworks.com/matlabcentral/fileexchange/51153-automatic-segmentation-of-brain-tumor-in-mr-images

http://www.mathworks.com/matlabcentral/fileexchange/44516-automatic-segmentation-of-dermoscopic-images

http://www.mathworks.com/matlabcentral/fileexchange/43518-graph-based-segmentation-of-retinal-layers-in-oct-images

http://www.mathworks.com/examples/image/mw/images_product-KMeansSegmentationExample-color-based-segmentation-using-k-means-clustering

http://www.mathworks.com/matlabcentral/fileexchange/53840-axon-segmentation

http://www.mathworks.com/matlabcentral/fileexchange/28330-a-fast-image-segmentation-using-delaunay-triangulation

http://www.mathworks.com/matlabcentral/fileexchange/31068-spatial-fuzzy-clustering-and-level-set-segmentation

Answer 2

[Image Analyst]

Are you kidding? There must be tens of thousands of them. Go here to find most of them: http://www.visionbib.com/bibliography/contents.html

Of course 99.99999% of those will not be applicable to your specific images, so it's best to show us your image and tell us what you want to measure.

[priya y]

Thank you sir for your reply. I am currently working on plant disease identification and classification ,so i want to segment the diseased part from the entire leaf (this method to be applied to multiple images at a time). I am currently using k-means clustering method,but it requires human intervention ie have to select the cluster containing the diseased part for every image.So i want an automated method for segmentation to apply to all images at a time.

[Image Analyst]

kmeans won't be any good if you think about it. If you can't figure out why, ask me.

Did you see where I said "so it's best to show us your image"?

[priya y]

i want to seperate the diseased part from the above attached images.

[Walter Roberson]

By the fall, the leaves on trees around here commonly look like the "blight" image. That does not necessarily indicate "disease", as there are plenty of insects around that eat leaves -- and as the leaves get ready to drop, the drying out process has to start somewhere on the leaf.

[Image Analyst]

Why not segment out the dark background and look at the average hue of what's left? You would need a reference color though because the colors may be normal for some plants. And we can't use kmeans because we don't know how many clusters there are, if any, and we need it to work for any amount of disease from 0% to 100%. We need the reference color because what if the leaves are dark purple? That color may indicate disease on a leaf that is normally green but would be perfectly fine for a plum tree. So you need to know what is "good" color and what is a deviation from that. You can use rgb2lab and compute the color difference delta E. And it's easier if you have nice photos, like a branch of leaves against black velvet. It would be harder of impossible if you were to just run around your neighborhood with a video camera pointing it around everywhere and expect it to find all the diseased leaves.

[Walter Roberson]

As the owner of a Double Flowering Plum (Prunus triloba 'Multiplex'), I regret to indicate that plum trees do not have dark purple leaves, not even in fall. We get some beautiful blooms, but the leaves stay completely green even through the first few frosts.

[Image Analyst]

Well, you're right. The plum trees we see are usually ornamental, rather than for their fruit, and they're purple. Here is one https://thegoodlifemarketplace.wordpress.com/fruit-trees/mini-orchard-set-up/plum-tree/

[Walter Roberson]

Looks like Prunus cerasifera, probably 'Krauter Vesuvius', and it appears to be part of the subgenus "prunus"; whereas it looks like the Prunus triloba might not be. But there is nontrivial dispute about what to call a plum and what not to...

So I wrote too quickly. Some plums do indeed have purple leaves -- but not even all of the ones recognized for their fruit.

Gee the Rosaceae (Rose family, which includes apples and a lot of the most important non-citrus fruit) are complicated!