how to calculate area of a feature in a gray scale image

I am having an Ultrasonic C-Scan gray scale image of a steel billet. I need to find the area of a feature in it by using intensity difference or texture difference or pattern recognition or any other suitable method. The main problem is that the boundries of feature are not demarkated clearly and intensity variation is very less compared to the surrounding area. Will be thankful if anybody is able to give some clues

Answer 1

[Image Analyst]

Other than how to upload an image about all I can say is that you can use image processing. You might also see my Image Segmentation Tutorial if you're a novice at image processing.

[sheethal sasidharan]

I am uploading a diagramatic representation ( http://imageshack.us/photo/my-images/197/diagramaticrepresentati.jpg/) along with original C-Scan image in gray (<http://imageshack.us/photo/my-images/7/cscanimageingrayscale.jpg/>) aswell as colour images (<http://imageshack.us/photo/my-images/442/45886590.jpg/)>. please go through the images a suggest a method for calculating area of equiaxed zone from either grey or colour scale

[Image Analyst]

I'd probably start with an edge detector and look at the direction of the gradients.

[sheethal sasidharan]

I have tried with edge detectors it will show multiple no pf edges in columnar zone, and there is not much difference in gradient, moreover some of the gray scale intensities in equiaxed zone are simmilar to that of columnar zone. Ref: http://imageshack.us/photo/my-images/7/cscanimageingrayscale.jpg/, http://imageshack.us/photo/my-images/442/45886590.jpg/