How to separate data sets/clusters that got saved in a single data file?

I have this follwing data. Col 1: Vertical Pixels, Col 2: Horizontal Pixels, Col 3: Index of 256x256 pixels, Col 4: Data 1, Col 5: Data 2. When I plot Col 3 and Col 5 (as shown in the figure) using 'imagesc', there appear two seprate clusters each containing a part of values from Col 5. My question is how do I extract the data or columns correponding to each cluster without visualizing it. Any leads will be highly appreciated. Many thanks.

9 77 19464 410575406 13

15 73 18446 410575406 11

17 74 18704 410575406 12

12 77 19467 410575406 17

9 76 19208 410575408 1

16 74 18703 410575406 13

11 77 19466 410575406 31

10 77 19465 410575406 21

16 73 18447 410575406 34

15 74 18702 410575406 80

Answer 1

[Cris LaPierre]

kmeans clustering algorithm?

data = [9	77	19464	410575406	13
15	73	18446	410575406	11
17	74	18704	410575406	12
12	77	19467	410575406	17
9	76	19208	410575408	1
16	74	18703	410575406	13
11	77	19466	410575406	31
10	77	19465	410575406	21
16	73	18447	410575406	34
15	74	18702	410575406	80]

data = 10×5

9 77 19464 410575406 13 15 73 18446 410575406 11 17 74 18704 410575406 12 12 77 19467 410575406 17 9 76 19208 410575408 1 16 74 18703 410575406 13 11 77 19466 410575406 31 10 77 19465 410575406 21 16 73 18447 410575406 34 15 74 18702 410575406 80

img = zeros(256);
img(data(:,3)) = data(:,5);
imagesc(img)
xlim([68 82])
ylim([7 20])

idx = kmeans(data(:,1:2),2);

figure
gscatter(data(:,2),data(:,1),idx)
axis image
axis ij
xlim([68 82])
ylim([7 20])

[Raju Kumar]

@Cris LaPierre Thanks for your reply. I think k=2 already indicates that we know the number of clusters (from the image). I would to like to do this clustering without visualizing the image (i.e. for unknown k).

[Cris LaPierre]

You could combine that with evalclusters to try to identify the ideal number of clusters.

data = [9	77	19464	410575406	13
15	73	18446	410575406	11
17	74	18704	410575406	12
12	77	19467	410575406	17
9	76	19208	410575408	1
16	74	18703	410575406	13
11	77	19466	410575406	31
10	77	19465	410575406	21
16	73	18447	410575406	34
15	74	18702	410575406	80]

data = 10×5

9 77 19464 410575406 13 15 73 18446 410575406 11 17 74 18704 410575406 12 12 77 19467 410575406 17 9 76 19208 410575408 1 16 74 18703 410575406 13 11 77 19466 410575406 31 10 77 19465 410575406 21 16 73 18447 410575406 34 15 74 18702 410575406 80

eva = evalclusters(data(:,1:2),'kmeans','CalinskiHarabasz','KList',1:6)

eva = 
  CalinskiHarabaszEvaluation with properties:

    NumObservations: 10
         InspectedK: [1 2 3 4 5 6]
    CriterionValues: [NaN 71.7241 65.8600 61.0545 49.7500 38.8343]
           OptimalK: 2

[Raju Kumar]

@Cris LaPierre Thanks but this finds more k even when there should not be any. I have found a way by applying logical thoughts looking at the data. But I appreciate your answers.

[Cris LaPierre]

Without having the actual data, it's hard to say for certain what would work. You have something working, wich is great. My next thought would be to look into some image processing techinques, like blob analysis perhaps?