Image Segmentation Prototyping and Evaluation Tool
An image before and after k-means = 16. (Source: Wikipedia)
The aim of this project is to create an interactive tool which shall allow users to load a sample image and experiment with a number of different segmentation algorithms. The user shall be able to choose which of the algorithms they wish to implement and then enter the variables that may be associated with the selected methods. Once the user has chosen the desired methods they can then draw around the area of the image they wish to be segmented. Through doing this a "golden standard" shall be created that the outputs given shall be compared against, showing the user the best method. Once completed the user can then chose to save the segmented image locally.
Information about Image Segmentation
Image segmentation is one of the most widely used image processing/analysis processes. Its goal is to divide an input image into distinct regions (sets of pixels known as superpixels), each of which corresponds to some meaningful object. Many algorithms and approaches have been proposed, making it difficult for non-specialists to select and make best use of the most appropriate one.
Image segmentation is incredibly useful in a wide variety of fields. For example image segmentation is increasingly being used in many clinical and research applications analysis. When the doctor has the x-ray image of the patient image segmentation can be used to clarify the image and determine what is wrong with the patient easier. Moreover it is also used in medicine with regards to locating tumours and studying the anatomical structure more thoroughly.
Image segmentation is also beginning to play a big part in security. With CCTV cameras not being that clear in the images they provide, image segmentation is a very useful method for manipulating the pixels in order to help clarify things like people’s faces. Similar things can be used for fingerprint and facial recognition.
Image segmentation is used a lot in geography. It is used a lot in human geography with regards to analysing an area from images of the town from a bird’s eye view. It can also be used for analysing physical geography a lot, such as viewing different layers of rock in cliffs.
There are some typical methods of image segmentation such as:
Thresholding: This method is based on a clip-level (or a threshold value) to turn a gray-scale image into a binary image.
Clustering: This method includes the k-means algorithm and segments pictures by calculating the distance to the set means.
Smoothing: In this technique an image is able to have some noise removed to make one of the algorithms more effective.
- Lakhvir Singh Sohal - Group Leader
- Alex Neves - Technical Lead
- Adam Hardacre - Site/Repository Master
- Nan Wang - GUI Designer
- Yuxuan Chen - Lead Programmer
- Jack Evans - Editor
- Tony Pridmore - Supervisor