Given the algorithm for recognizing images that need to be implemented - https://docs.google.com/file/d/0BxxxoAIFa58na1J3Q1pTV1NFRjQ/edit

The problem is that there is practically no information on these algorithms. I guess they are just called differently.

Tell me, what kind of algorithms and in what direction to dig.

  • What you do not understand in this algorithm? Remotely resembles the k-means method. You can look in the direction of the Rosenblatt perceptron. - hedgehogues

1 answer 1

Here are the variants of the ISODATA algorithm. When initially set the number of classes. Also some algorithms are possible k-means. The SPEKTR algorithm is similar to the rearrangement algorithm after clustering using quality functionals (dispersion-covariance criteria)