Color image Segmentation in Optical Microscopy

The goal of this semester project is to build a software which provides segmentations methods for any kind of images. The biologists produces many color images in histology with optical microscopy and they need to segment them before they can quantify some parts of the images. Most of the time, the biologists have to find thresholds manually in order to have well segmented images. Our software, a plugin designed for ImageJ, provides algorithms which can segment images automatically and can do it for any number of images all at once. The user can interact with the plugin by giving the initialization of the algorithms. The algorithms which have been chosen to perform that task are the K-Means and the Hidden Markov Model. Our software also provide s a tool which allows the user to transform segmented images into binary images which can be used to analyze some spatial properties.

Orginal Image

Segmented image

Binarized image