MATLAB Answers

0

Area detection and measurement

Asked by Jack Lovell on 22 Jul 2019
Latest activity Answered by Yogesh Khurana on 2 Aug 2019
Hi all,
I have a large dataset of images which consists of .AWG files of human arms, my goal is to be able to measure a red area of skin (allergy test) for each photo. The photos were taken at the same distance from the arm, although the arms aren't consistent in the left to right plane in the photo (some are fuirther from center than others, maybe an issue?). I was wondering if there was a way to automate this problem use a genral script that has been written as a starting point as I have never done image analysis before. I am currently working with ImageAnalyst's Blobdemo() script, but was wdonering if there would be abetter method as the photos are all in RGB, and the contrast of white skin to red would allow for edge detection to be much easier.
Thank you for your help!

  1 Comment

David K.
on 22 Jul 2019
I am definitely not as experienced as ImageAnalyst is, but very often one of the first steps when working with color is to convert to greyscale or black/white. If the red is actually consistently more red than any other part of the image, you could start by finding out if the red of each pixel is above a certain threshold that you specified. If the value is not consistent enough there are ways to use an adaptive threshold that changes based off of nearby pixels.
With the thereshold applied or even just using the R values in the conversion to greyscale, your image should be closer to the example setup in the demo.
Hope this helps

Sign in to comment.

1 Answer

Answer by Yogesh Khurana on 2 Aug 2019

There are different ways by which you can detect edges which will help in this situation. You can convert the image to GrayScale using rgb2gray or convert to hsv image using rgb2hsv. Apply edge detection techniques according to your constraints after converting the image to grayscale or hsv.
I am providing you some links where you can explore more about this:
Hope this helps!

  0 Comments

Sign in to comment.