Implementation of the Enhanced LBG Algorithm Based on the paper "Neural Networks 14:1219-1237" that can be found in http://citeseer.ist.psu.edu/patan01enhanced.html .

Parameters

points Input points.

dim Dimension of the points.

numpoints Num of points in **points.

codebook Pointer to the output codebook. Must be allocated.

numCB Number of points in the codebook.

num_steps The maximum number of steps. One step is already a good compromise between time and quality.

closest_cb Return the closest codebook to each point. Must be allocated.

rand_state A random number generator state. Should be already initialized by av_lfg_init().

Returns

< 0 in case of error, 0 otherwise

Definition at line 371 of file elbg.c.

Referenced by a64multi_encode_frame(), avpriv_init_elbg(), encode_frame(), filter_frame(), generate_codebook(), and quantize().

Initialize the **codebook vector for the elbg algorithm.

If you have already a codebook and you want to refine it, you shouldn't call this function. If numpoints < 8*numCB this function fills **codebook with random numbers. If not, it calls avpriv_do_elbg for a (smaller) random sample of the points in **points. Get the same parameters as avpriv_do_elbg.

Returns

< 0 in case of error, 0 otherwise

Definition at line 337 of file elbg.c.

Referenced by a64multi_encode_frame(), avpriv_init_elbg(), encode_frame(), filter_frame(), generate_codebook(), and quantize().