Region Growing 

This was a common problem in the book which I solved ... Yea I guess your argument is that it should be -1 Acctually the magnitude is compared ... the sign does not matter .. sorry for the confusion .. i forgot to mention that ..

Actually we are taking only the magnitudes. i.e | f(x,y)* - f(x,y)|

Thank you so much.. I have made playlists of my videos please visit my channel for the same .. thank you 😊👍

Thank you so much for the feedback ..😊👍

Gowher Malla yes acctually the condition of greater than or greater than equal to can depend on the program or the developer Acctually the pixel values have a range of 1 to 255 hence the greater than or greater than equal to does not have a significant difference

Deepak Mishra thank you so very much ..... these words really mean a lot .... ya i will complete the image processing course soon ... thank you so much ...

Srinidhi Bhat thank you so very much for your valuable feedback i will definately look to these points and accordingly make the next video ... Thank you so very much ..

+nandana 97 most welcome ..😊

Thank you so much 😊

Thank you so very much for the feedback .. 👍😊

Thanks for the feedback

Trigger 7 thank you so very much .....

ankit most welcome ..

Gowher Malla sorry i did not get your question ... After selecting the seed point we take the differnce of the current pixel with the seed point and compare with threshold

Then the 3s thay are on the borders will also be included in the region a.

@@pratikian So in total how many isolated pixels will be there?

Ohh you are asking with respect to region growing algorithm ... Yeh for 4 connected pixels you will only choose those points in 4 neighbours .. This that i have done in the video is 8 connected since i have considered all the 8 neighbours Yes the answer will be differnt

Gowher Malla ohh acha we take the magnitude. Absolute difference

Gowher Malla well that totally depends on how you code but i guess region growing says there should be 1 threshold ..

Yadnesh Khode all the very best bhai ... 👍

Kal semester hai ab😂😂

Choose a number that is in the middle of the range, here the range is 0-7, so you can choose 3 or 4.

@@pratikian Thank you, and for the values of each pixel we want to segment, can you explain where we can build them from?

@@imanesamiry1114 so the values of the pixels of the image will be given to you. If you are practically implementing it where you are given an image, you can use the imread function of Matlab or opencv library of python to get the values

@@pratikian thank you for your explanations :)

+Gowher Malla most welcome ..

So in real life situations the threshold acctually depends on how you want to segment the image. If its a sum for the exam they will give you the threshold . If it is not mentioned you can assume a reasonable threshold that is somewhere in the middle of the grayscale values.

Thank you bro

Oblivion Dew do you mean that after taking the difference u get 3 that is when you are comapring with the threshold ... here diffenent books say different things so acctually it is totally your choice whether you want to consider the equal to sign or not ..

Thanks for replying What i meant is that the seed point that im finding its neighbors is equal to the threshold If the seed is 3 and the threshold is 3 Will there be a difference in the solution ?

Oblivion Dew ohh you are telling when the seed is 3 well there can be a slight change in the sollution it is not quite advisable the best seeds should be either 6,7,1 or 2

Pratik Jain thanks so much In my college the prof in final exam Made the seed =3 and threshold =3 It s so confusing

Oblivion Dew well it is not at all wrong see let me explain you in the real world applications this algorithm was made to basically segment the objects from a picture Consuder a picture of a ball say so if i want to only select the pixels of the ball i will select a seen point from inside the ball .. now in real world there is 256 levels 256 values where choosing a seed poibt such questions would never arise but this sum is only for our understanding for purpose if ease 7 levels are taken but in exams especially mum university the papers are checked according to some sollution hence a better practice is to mention clearly what ever you have used

Sure .. thank you for the suggestion 😅

I did not get you ?

The sum that I have talked about in the video has pixel values only from 0 to 7 (3 bit image) so if you keep threshold 100 no matter what image you have the whole image will be in one segment. For a 8 bit standard image you can have threshold 100 but still it is very much likely that you wont get good results but nevertheless thresholds are application depedent.

Start the same process on both the seeds Same algorithm just 2 start points

@@pratikian Using 1 seed, the parts of the image included the region had been marked as 'a' by you. In case of 2 seeds should i also check those regions, marked as 'a', for selecting region for second seed?

@@pranavvenkatesan9712 acctually the main aim of selecting 2 seed points is that you are going to have to different regions ... So when you select 2 seed points ... you will have seeds which differ in large magnitude ... Like over here In 1st eg I took seed as 6 so other seed that i would prefer would be 1 or 0 That is when we do it practically Well to answer your question if the threshold satisfies then you should also check for all the a also (that belongs to the previous region

@@pratikian ya then wouldnt the second seed remove the region formed by the first seed since the difference may not satisfy with the threshold value?

@@pranavvenkatesan9712 no basically you are getting 2 regions Say you mark the 2nd seed with b So you are getting 2 regions And its just that both will be overlapping.