07-01-2013 06:58 AM
Hi alex,
I had tried it.Output from your's color mask file is better than mine.when class 1 is given as input.
Thank you
07-01-2013 07:50 AM
Madhubalan,
please mark your thread as solved and say 'thank you' with kudos
Regards,
Alex
07-01-2013 09:10 AM
Hi alex,
can you please create percentage determination of the dominant color in your color mask vi file.
Do you have any source code for object detection?.
Thank you.
07-01-2013 11:58 AM
ok, I've to admit "percentage determination of the dominant color" might not be that trivial as I first thought, re-reading your question more carefully.
why?
the k-means algorithm gives you a centroid,
whose specific color - due to averaging - mustn't necessarialy be part of the original color ramp.
what do you "percentage of dominant color" expect to be?
Do you have any source code for object detection?.
this is a completly different topic; what do you consider an object to be?
it looks like you do have the vision toolkit; object detection should be feasible with this toolkit. but I can't tell you how.
You should post another thread for this new topic!
07-01-2013 02:37 PM
07-02-2013 04:32 AM
@Madhubalan wrote:
Hi Alex ,
In the color output,we are getting a color,for example,the file which I had sent previously(attached file,color percentage),in which pink is the dominant color of the input image,I would like to know how much percentage of pink color present in the image.
Object is a printer to be detected from an image( I.e printer has to be detected when placed in a shelf or on a table)
Or even screw from the input image I had attached in the previous post.
Thank you
listen,
there is not just ONE pink color in the image, there are a few hundred different shades of pinkish-red color.
k-means does NOT give you a scattering parameter
k-means gives you a centroid or mean which is a location parameter
In the following picture, µ is the location parameter and σ is the scatter parameter
07-02-2013 06:27 AM
Hi Alex,
Can you please send the source code for that graph,it will be very helpful for me.
Thank you
07-02-2013 10:32 AM - edited 07-02-2013 10:34 AM
hi,
the picture in the post above was from a wikipedia-entry about normal distribution
the easiest way to create a histogramgraph in labview would be to use
http://zone.ni.com/reference/en-XX/help/371361J-01/gmath/histogram_discrete/
or the mor generic
http://zone.ni.com/reference/en-XX/help/371361H-01/gmath/general_histogram_vi/
from the Probability & Statistics VIs palette
07-03-2013 07:52 AM
Hi Alex,
Your code works very very fine.Thank you so much.
How to perform color histogram matching((matching the object from the original input image with the obtained binary(output) image)) like in this picture((example Color histogram matching attached)) on your color mask.vi file.
I have attached my input file image.
Thank you.
07-06-2013 07:59 AM - edited 07-06-2013 07:59 AM
Please define "matching";
do you want to hi-light certain areas on your input image via superposition of an overlay?