Color Thresholding

 

Home
Contrast Stretching
LoG Filter
Periodic Noise
Binary Methods
Particle Counting
Color Thresholding

Apr 23, 2001                                        Lab 6

 

Thresholding is a key element is selecting information from data, important material from background material in images.  Much of the thresholding done so far has involved setting boundaries based on gray values or intensities of image pixels.  In this lab, the thresholding is to be done based on color values.  Depending on the image characteristics and what analysis is desired, the thresholding might be done either in RGB or in HSI space.

In the following picture, what is the number of yellow circles and what is their average area?  What about red circles?

A more difficult task is to change the color of certain parts of an image.  In the picture of the two birds, how would you change the shoulder part of the wing on the bird on the right from green to yellow?  Remember that the purpose of this lab is pedagogical, not aesthetic, i.e. don't spend too much time trying to get a perfect match along border pixels.  It is the procedure that is important.  

Sometimes the selection of a color range requires using a Boolean condition between the different planes of a color image.  Looking at the three-plane montage of the following picture in both the RGB and the HSI domain, decide which is the better domain and what combination of criteria will most easily allow the separation of the red from the pink blocks.  What about distinguishing the red from the brown blocks?

 

Last modified on April 23, 2001