How to measure an angle and keeping the one that verify a certain measure defined priviously ?

I have a network and i want to measure the angle connecting every two vectors defined priviously

by coordinates(feature points detected) and passing from angle to another with keeping angles

that verify a condition of measure

below the network with mentioned coordinates in red :

A far purpose is to keep form that verify X, Y, L, T and -

X : Four angles of 90 degrees

Y : One angles <90 and two others >90

L : One angle about 90 and the other about 270

- : 180

How can i proceed

any ideas please!

thanks

[mika]

Any ideas please!

I used TSP_NN to find a way to connect Feature points detected priviously

But what i'm looking for is to measure angles between one points and the two next to identify what template it present.

X template : Four angles of 90 degrees

Y template : One angles <90 and two others >90

L template : One angle about 90 and the other about 270

- template : 180

Help is much appreciated

thanx

[mika]

Can any one please help me doing this

I'm starting with matlab.

[Geoff Hayes]

Perhaps start by writing out the pseudo code for the function that will solve the above problem. Do you already have the list of feature points and now wish to just find angles between every consecutive triplet?

[mika]

Yes i have the list of feature points, but don't know how to proceed to measure angles.

[Geoff Hayes]

What are the properties for a single feature point? Is there a coordinate associated with it?

[mika]

it's only a 148*2 matrice below an example

ending_pts =

     3   791
     8   388
    10   570
    17   285
    24   836
    32   342
    35   656
    43   339
    44   634
    53   656
    55   445
    57   657
    59   179
    60    62
    62   217
    64   618
    73   629
    74   629
    77   617
    84   867
   126   401
   133   920
   138    82
   141   386
   158   528
   168    65
   184   660
   191   650
   198    13
   208   537
   219   730
   224   524
   227   527
   230   957
   233   955
   236   954
   253    89
   253   917
   258   964
   263   836
   275   478
   285   622
   289   240
   292   418
   295   747
   297   748
   297   946
   299   782
   311   395
   317   777
   318   105
   322   166
   326   792
   355   927
   360   103
   372   853
   384   110
   387   534
   387   574
   391   421
   398   579
   404   702
   419   324
   422   323
   424   374
   425   908
   435   101
   443    30
   464   761
   468   899
   470   897
   471   494
   475   754
   488   888
   496   707
   496   799
   505   565
   506   567
   513   551
   517   878
   523   531
   528   460
   542   187
   555   210
   558   449
   572   404
   576   403
   577   597
   583   397
   588   711
   592   245
   595   873
   596    83
   621     4
   621   900
   627   692
   650   559
   654   558
   656   870
   663    83
   665   444
   665   665
   666   412
   669   155
   669   552
   672   361
   673   359
   673   485
   680   660
   685   486
   694   352
   695   711
   698   650
   709   231
   712   624
   728   435
   731   891
   732   309
   733    81
   737   421
   737   425
   742    53
   743    51
   746   839
   747   842
   752   347
   753    84
   755   879
   771     9
   776   328
   778   395
   829   338
   829   340
   836   333
   853   324
   863   370
   913   581
   917   644
   926   455
   936   396
   943   741
   949   609
   949   620
   954   332
   959   169
   966   329
   967   327
   972   725

[Geoff Hayes]

You posted this same set of data at http://www.mathworks.com/matlabcentral/answers/131183-error-using-reshape-size-arguments-must-be-integer-scalars. Are these all the feature points? Are these (x,y) coordinates? The first column is in ascending order - is this important? What does this data set mean?

[mika]

yes these are (x,y) coordinates of all the feature points

this data set is a the ending points of a skeleton extracted using Skeleton Endpoints

not important the order

[Geoff Hayes]

I would have thought that order would be important… You can use atan2 or atan2d to determine the angle between two feature points (coordinates). Just type help atan2d in the command window for details.

[mika]

i have to determine the angle between three feature points not two and each time i change two feature points until finding the required measure of angle

[Geoff Hayes]

Please clarify - you have three distinct points and you want to find one angle?

[mika]

like mentioned in the image below three points are required and each time changing one feature until reaching the required angle measure

[Geoff Hayes]

Ok…so how from the list above do you decide to group any three feature points? What is the logic for that?

[mika]

I'm still lookig for a logic i thinked about the nearest neighbor but i don't know how to implemet it

[Geoff Hayes]

Do you have some sort of cost function?

[mika]

no but may be i can use a cost function to switch from segment to another ??

[mika]

No idea ??

please reply. thx.

[Geoff Hayes]

Have you decided on how to group any three feature points?

[mika]

Sorry for the late reply i taked a time to decide how to group feature points.

I will consider every group of Feature points as a cell defined priviously by a Voronoi algorithm to extract skeleton below from initial image.

Knowing that this cell have a centroid equidistant from all the boundaries which can lead to estimate the missing segment of that cell.

Any ideas to approximate segments.

thx