void robot_mt_cell (FILTER_DESC *filter_desc)
{
NEURON_LAYER_LIST *n_list;
NEURON_LAYER *nl_complex_left, *nl_complex_right, *nl_complex_binocular;
PARAM_LIST *p_list;
int i;
//int j;
int num_neurons, numNL, nNumParam;
float fltComplexBinocular, fltComplexLeft, fltComplexRight, k;
for (numNL = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, numNL++)
;
if (numNL != 3)
{
Erro ("Wrong number of neuron layers. The robot_mt_cell must be applied on three neuron layers.", "", "");
return;
}
for (i = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, i++)
{
if (n_list->next != NULL)
{
if (n_list->neuron_layer->output_type != n_list->next->neuron_layer->output_type)
{
Erro ("The robot_mt_cell must be applied on neuron layers with the same output_type.", "", "");
return;
}
}
}
// Achar o numero de parametros
for (nNumParam = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, nNumParam++);
nNumParam--;
// Numero de neuron layers deve ser igual a tres
if (nNumParam != 1)
{
Erro ("The robot_sum_filter_normalized must have one parameter: Selectivity.", "", "");
return;
}
// Seletividade
k = filter_desc->filter_params->next->param.fval;
nl_complex_left = filter_desc->neuron_layer_list->neuron_layer;
nl_complex_right = filter_desc->neuron_layer_list->next->neuron_layer;
nl_complex_binocular = filter_desc->neuron_layer_list->next->next->neuron_layer;
num_neurons = get_num_neurons (filter_desc->output->dimentions);
for (i = 0; i < num_neurons; i++)
{
fltComplexLeft = nl_complex_left->neuron_vector[i].output.fval;
fltComplexRight = nl_complex_right->neuron_vector[i].output.fval;
fltComplexBinocular = nl_complex_binocular->neuron_vector[i].output.fval;
filter_desc->output->neuron_vector[i].output.fval = fltComplexBinocular / (fltComplexLeft + fltComplexRight + k);
}
}
void ComandoCall::Interpreta( Contexto& C )
{
if ( (!identificador) || (!param_reais) )
throw Erro("\nUnexpected NULL pointer.");
/* Obtém Procedimento */
Procedimento * P = C.obtemProcedimento(*identificador);
/* Checa número de Parâmetros */
ListaVariaveis * params = P->obtemParametros();
if ( params->tamanho() != param_reais->tamanho() )
throw Erro("\nProcedimento inválido: número de parâmetros errado");
/* Atribui parâmetros e adiciona as variáveis no RA */
int i;
std::vector<double> p_reais = param_reais->Avalia(C);
/* Adiciona novo Registro de Ativação */
C.adicionaRA();
for (i = 0; i < params->tamanho() ; i++)
{
C.adicionaVariavel((*params)[i], p_reais[i]);
}
/* Executa Procedimento */
Comando * comandoLocal = P->obtemLocal();
comandoLocal->Interpreta(C);
ListaComandos * cmds = P->obtemComandos();
cmds->Interpreta(C);
/* Remove Registro de Ativação */
C.removeRA();
}
//---------------------------------------------------------------------------
bool __fastcall TRawPrint::FechaDispositivo(void)
{
bool Retorno = true;
try
{
if (DispAberto)
{
if (PaginaAberta)
{
if(!EndPagePrinter(hPrinter))
Erro("Erro ao Finalizar Página");
if(!EndDocPrinter(hPrinter))
Erro("Erro ao Finalizar Documento");
else
{
if(!ClosePrinter(hPrinter))
{
Erro("Erro ao Fechar a Impressora");
Retorno = false;
}
DispAberto = false;
}
}
}
}
catch(Exception & E)
{
Erro("Erro ao Fechar a Impressora.\r\n\r\n" + E.Message + "\r\n");
}
return Retorno;
};
void
hsv_v_filter(FILTER_DESC *filter_desc)
{
// http://en.wikipedia.org/wiki/HSL_and_HSV
// http://cs.haifa.ac.il/hagit/courses/ist/Lectures/Demos/ColorApplet2/t_convert.html
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
int xo, yo, wi, hi, wo, ho;
int r, g, b;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 1)
{
Erro ("Error: Wrong number of parameters. The blue_mask_filter must have no parameters.", "", "");
return;
}
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
// Input-Output width
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
if (wi != wo || hi != ho)
{
Erro ("Error: Input and output layers on blue_channel_mask filter must have the same 2D size.", "", "");
return;
}
for (yo = 0; yo < ho; yo++)
{
for (xo = 0; xo < wo; xo++)
{
r = RED(nl_input->neuron_vector[xo + yo * wo].output.ival);
g = GREEN(nl_input->neuron_vector[xo + yo * wo].output.ival);
b = BLUE(nl_input->neuron_vector[xo + yo * wo].output.ival);
nl_output->neuron_vector[xo + yo * wo].output.ival = max_value(r, g, b);
}
}
}
void robot_complex_cell (FILTER_DESC *filter_desc)
{
NEURON_LAYER_LIST *n_list;
NEURON_LAYER **nl;
int i, j;
int num_neurons, numNL;
double dif_accumulator, a, b;
for (numNL = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, numNL++)
;
if (filter_desc->private_state == NULL)
{
if ((numNL % 2) != 0)
{
Erro ("Wrong number of neuron layers. The robot_complex_cell must be applied on a even number of neuron layers.", "", "");
return;
}
for (i = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, i++)
{
if (n_list->next != NULL)
{
if (n_list->neuron_layer->output_type != n_list->next->neuron_layer->output_type)
{
Erro ("The robot_complex_cell must be applied on neuron layers with the same output_type.", "", "");
return;
}
}
}
nl = (NEURON_LAYER**)malloc(sizeof(NEURON_LAYER*)*numNL);
for (i = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, i++)
nl[i] = n_list->neuron_layer;
filter_desc->private_state = (void*)nl;
}
else
{
nl = (NEURON_LAYER**)filter_desc->private_state;
}
num_neurons = get_num_neurons (filter_desc->output->dimentions);
for (i = 0; i < num_neurons; i++)
{
dif_accumulator = 0.0;
for (j = 0; j < (numNL / 2); j++)
{
a = nl[j]->neuron_vector[i].output.fval;
b = nl[j + (numNL / 2)]->neuron_vector[i].output.fval;
dif_accumulator += (a*a + b*b);
}
filter_desc->output->neuron_vector[i].output.fval = dif_accumulator;
}
}
void
Cr_filter(FILTER_DESC *filter_desc)
{
// http://www.equasys.de/colorconversion.html
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
int xo, yo, wi, hi, wo, ho;
double r, g, b;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 1)
{
Erro ("Error: Wrong number of parameters. The blue_mask_filter must have no parameters.", "", "");
return;
}
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
// Input-Output width
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
if (wi != wo || hi != ho)
{
Erro ("Error: Input and output layers on blue_channel_mask filter must have the same 2D size.", "", "");
return;
}
for (yo = 0; yo < ho; yo++)
{
for (xo = 0; xo < wo; xo++)
{
r = (double) RED(nl_input->neuron_vector[xo + yo * wo].output.ival);
g = (double) GREEN(nl_input->neuron_vector[xo + yo * wo].output.ival);
b = (double) BLUE(nl_input->neuron_vector[xo + yo * wo].output.ival);
nl_output->neuron_vector[xo + yo * wo].output.ival = (int) (0.500 * r - 0.419 * g - 0.081 * b + 128.0);
}
}
}
void calculate_error_function_from_trained_output_filter(FILTER_DESC *filter_desc) {
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
int wi, hi, xo, yo, wo, ho;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 1) {
Erro("Error: Wrong number of parameters. The calculate_error_function_from_trained_output_filter must have no parameters.", "", "");
return;
}
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
// Input Layer Dimensions
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
// Output Layer Dimensions
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
// Layer size verification
if (wo != wi || ho != hi) {
Erro("Error: calculate_error_function_from_trained_output_filter input and output layers must have te Same Size.", "", "");
return;
}
// Checks whether the neuron layer has been trained
if (!gaussian_filtered_training_pattern) {
return;
}
for (yo = 0; yo < ho; yo++) {
for (xo = 0; xo < wo; xo++) {
// Error = Reference - Output
// Simple Difference calculate (other error functions could be also applied)
nl_output->neuron_vector[xo + yo * wo].output.fval = gaussian_filtered_training_pattern[xo + yo * wo] - nl_input->neuron_vector[xo + yo * wo].output.fval;
//nl_output->neuron_vector[xo + yo * wo].output = nl_input->neuron_vector[xi + yi * wi].output;
}
}
}
void make_input_image_class_cnae (INPUT_DESC *input, int w, int h)
{
char message[256];
input->tfw = nearest_power_of_2 (w);
input->tfh = nearest_power_of_2 (h);
input->ww = w;
input->wh = h;
switch(TYPE_SHOW)
{
case SHOW_FRAME:
input->vpw = input->neuron_layer->dimentions.x;
input->vph = input->neuron_layer->dimentions.y;
break;
case SHOW_WINDOW:
input->vph = h;
input->vpw = w;
break;
default:
sprintf(message,"%d. It can be SHOW_FRAME or SHOW_WINDOW.",TYPE_SHOW);
Erro ("Invalid Type Show ", message, " Error in update_input_image.");
return;
}
input->vpxo = 0;
input->vpyo = h - input->vph;
if(input->image == NULL)
input->image = (GLubyte *) alloc_mem (input->tfw * input->tfh * 3 * sizeof (GLubyte));
}
void ordena(linha_t **lin, unsigned int num){
if(!num)
Erro("Nao existe campo 0");
*lin = ordena_linha(lin, NULL, num);
}
//---------------------------------------------------------------------------
bool __fastcall TRawPrint::ValidaComando(AnsiString Comando)
{
bool Retorno;
AnsiString Parametro;
int Indice = 1;
int Valor;
try
{
for(int i = 0; i < Comando.Length() / 3; i++)//Percorre o Comando
{
Parametro = Comando.SubString(Indice, 3);
if(Indice < 4)
Indice = Indice + 3 + i;
else
Indice = Indice + 3;//Índice de Início do Último Parâmetro (7)
Valor = StrToIntDef(Parametro, -1); //Verifica se é Inteiro
if(Valor >= 0 && Valor < 256) //Verifica se o Valor do Parâmetro está de acordo com Manual da Epson
{
Retorno = true;
}
else
{
Retorno = false;
}
}
}
catch(Exception & E)
{
Erro("Erro ao validar Comando informado no Texto.\r\n\r\n" + E.Message + "\r\n");
}
return Retorno;
};
void blue_mask_filter (FILTER_DESC *filter_desc)
{
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
int xi, yi, wi, hi, wo, ho;
//int color_channel;
//unsigned int (*mask_func)(unsigned int);
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 1)
{
Erro ("Error: Wrong number of parameters. The blue_mask_filter must have no parameters.", "", "");
return;
}
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
// Input-Output width
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
if(wi != wo || hi != ho)
{
Erro ("Error: Input and output layers on blue_channel_mask filter must have the same 2D size.", "", "");
return;
}
for (yi = 0; yi < hi; yi++)
for(xi = 0; xi < wi; xi++)
nl_output->neuron_vector[xi + yi * wi].output.ival = BLUE(nl_input->neuron_vector[xi + yi * wi].output.ival);
}
void scale_nl_filter(FILTER_DESC *filter_desc) {
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
float scale_factor, k;
int xi, yi, wi, hi, xo, yo, wo, ho;
//int delta_xo,delta_yo,corrected_xo,corrected_yo; //Center offset for image scale
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 4) {
Erro("Error: Wrong number of parameters. The scale_nl_filter must have only three parameter <scale_factor> <delta_xo> <delta_yo>.", "", "");
return;
}
// Gets the Filter Parameters
scale_factor = filter_desc->filter_params->next->param.fval;
//delta_xo = filter_desc->filter_params->next->next->param.ival;
//delta_yo = filter_desc->filter_params->next->next->next->param.ival;
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
// Calculates the scaling constant
k = 1.0 / scale_factor;
//#ifdef _OPENMP
// #pragma omp parallel for private(yo,yi,xo,xi)
//#endif
for (yo = 0; yo < ho; yo++) {
yi = (int) (k * (float) yo + .5f); //+ (int)(scale_factor*delta_yo);
for (xo = 0; xo < wo; xo++) {
xi = (int) (k * (float) xo + .5f); //+ (int)(scale_factor*delta_xo);
if ((xi >= 0) && (xi < wi) && (yi >= 0) && (yi < hi))
nl_output->neuron_vector[xo + yo * wo].output = nl_input->neuron_vector[xi + yi * wi].output;
else
nl_output->neuron_vector[xo + yo * wo].output.ival = 0;
}
}
}
void rotate_nl_filter(FILTER_DESC *filter_desc) {
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
float angle, cos_angle, sin_angle;
int xi, yi, wi, hi, xo, yo, wo, ho;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 2) {
Erro("Error: Wrong number of parameters. The rotate_nl_filter must have only one parameter <angle>.", "", "");
return;
}
// Gets the Filter Parameters
angle = filter_desc->filter_params->next->param.fval;
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
angle *= pi / 180.0f;
cos_angle = cos(angle);
sin_angle = sin(angle);
//#ifdef _OPENMP
// #pragma omp parallel for private(yo,yi,xo,xi)
//#endif
for (yo = 0; yo < ho; yo++) {
for (xo = 0; xo < wo; xo++) {
xi = (int) (cos_angle * (float) xo + sin_angle * (float) yo + .5f);
yi = (int) (-sin_angle * (float) xo + cos_angle * (float) yo + .5f);
if ((xi >= 0) && (xi < wi) && (yi >= 0) && (yi < hi))
nl_output->neuron_vector[xo + yo * wo].output = nl_input->neuron_vector[xi + yi * wi].output;
else
nl_output->neuron_vector[xo + yo * wo].output.ival = 0;
}
}
}
void ListaProcedimentos::AdicionaProcedimento(Procedimento * P) {
std::pair<std::string, Procedimento *> new_proc(P->obtemNome(), P);
bool OK = false;
OK = this->procedimentos_do_prog->insert(new_proc).second;
if (OK) return;
else throw Erro("\nProcedimento declarado duas vezes"+P->obtemNome());
}
void translate_nl_filter(FILTER_DESC *filter_desc) {
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
float x_offset, y_offset;
int xi, yi, wi, hi, xo, yo, wo, ho;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 3) {
Erro("Error: Wrong number of parameters. The rotate_nl_filter must have two parameters <x_offset> <y_offset>.", "", "");
return;
}
// Gets the Filter Parameters - The Pointers And The Values - Void pointers must be casted
x_offset = *((float*) (filter_desc->filter_params->next->param.pval));
y_offset = *((float*) (filter_desc->filter_params->next->next->param.pval));
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
// Parallel translation filter capabilities where OpenMP available
//#ifdef _OPENMP
// #pragma omp parallel for private(yo,yi,xo,xi)
//#endif
for (yo = 0; yo < ho; yo++) {
yi = (int) ((float) yo + y_offset + .5f);
for (xo = 0; xo < wo; xo++) {
xi = (int) ((float) xo + x_offset + .5f);
if ((xi >= 0) && (xi < wi) && (yi >= 0) && (yi < hi))
nl_output->neuron_vector[xo + yo * wo].output = nl_input->neuron_vector[xi + yi * wi].output;
else
nl_output->neuron_vector[xo + yo * wo].output.ival = 0;
}
}
}
//---------------------------------------------------------------------------
bool __fastcall TRawPrint::DispositivoAberto(void)
{
const AnsiString cRaw = "RAW";
_DOC_INFO_1A DocInfo;
bool Retorno;
Retorno = OpenPrinter(PChar(NomeDispositivo.c_str()), &hPrinter, NULL);
try
{
if(Retorno)
{
DocInfo.pDocName = NomeTrabalho.c_str();
DocInfo.pOutputFile = PChar(char(0));
DocInfo.pDatatype = PChar(char(0));
if (StartDocPrinter(hPrinter, 1, (LPBYTE)&DocInfo) == 0)
{
Erro("Erro ao iniciar Documento para Impressora");
ClosePrinter(hPrinter);
Retorno = false;
}
else
{
DispAberto = True;
NovaPagina();
Retorno = true;
}
}
else
{
Erro("Erro ao abrir a Impressora: " + NomeDispositivo);
Retorno = false;
}
}
catch(Exception & E)
{
Erro("Erro ao abrir a Impressora: " + NomeDispositivo + ".\r\n\r\n" + E.Message + "\r\n");
}
return(Retorno);
};
void
scale_and_translate_filter(FILTER_DESC *filter_desc)
{
NEURON_LAYER_LIST *n_list = NULL;
PARAM_LIST *p_list = NULL;
NEURON_LAYER *nl_input = NULL;
int i, ho, wo, hi, wi, xo, yo, xi, yi;
float scale_factor_x, scale_factor_y;
// Checks Neuron Layers
for (i = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, i++)
;
if (i != 1)
{
Erro ("Wrong number of neuron layers. Scale filter must be applied on only one neuron layer.", "", "");
return;
}
// Gets the Inputs Neuron Layers
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Check Parameters
for (i = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, i++)
;
// Gets the Input Neuron Layer Dimentions
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
// Gets the Output Neuron Layer Dimentions
wo = filter_desc->output->dimentions.x;
ho = filter_desc->output->dimentions.y;
scale_factor_x = (float) wi / (float) wo;
scale_factor_y = (float) hi / (float) ho;
for (xo = 0; xo < wo; xo++)
{
xi = scale_factor_x * xo + g_translate_x;
for (yo = 0; yo < ho; yo++)
{
yi = scale_factor_y * yo + g_translate_y;
if ((xi >= 0) && (xi < wi) && (yi >= 0) && (yi < hi))
filter_desc->output->neuron_vector[(yo * wo) + xo].output = nl_input->neuron_vector[(yi * wi) + xi].output;
else
filter_desc->output->neuron_vector[(yo * wo) + xo].output.ival = 0;
}
}
}
void
make_input_image_character(INPUT_DESC *input)
{
int i;
char message[256];
int w, h;
w = input->neuron_layer->dimentions.x;
h = input->neuron_layer->dimentions.y;
// Computes the input image dimentions
input->tfw = nearest_power_of_2(w);
input->tfh = nearest_power_of_2(h);
// Saves the image dimentions
input->ww = w;
input->wh = h;
// Computes the input image dimentions
input->tfw = nearest_power_of_2(w);
input->tfh = nearest_power_of_2(h);
switch (TYPE_SHOW)
{
case SHOW_FRAME:
input->vpw = input->neuron_layer->dimentions.x;
input->vph = input->neuron_layer->dimentions.y;
break;
case SHOW_WINDOW:
input->vpw = input->ww;
input->vph = input->wh;
break;
default:
sprintf(message,"%d. It can be SHOW_FRAME or SHOW_WINDOW.",TYPE_SHOW);
Erro("Invalid Type Show ", message, " Error in update_input_image.");
return;
}
input->vpxo = 0;
input->vpyo = w - input->vph;
if (input->image == (GLubyte *) NULL)
{
input->image = (GLubyte *) alloc_mem (3 * input->tfw * input->tfh * sizeof(GLubyte));
for (i = 0; i < input->tfh * input->tfw * 3; i++)
input->image[i] = 0;
}
return;
}
/*
*********************************************************************************
* Function: translate_nl_filter *
* Description: *
* Inputs: *
* Output: *
*********************************************************************************
*/
void gaussian_nl_filter(FILTER_DESC *filter_desc) {
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
//float x_offset, y_offset;
//int xi, yi, wi, hi, xo, yo;
int wi, hi;
int i;
// float* input_image_r = NULL, *input_image_g = NULL, *input_image_b = NULL;
float *output_image_r = NULL, *output_image_g = NULL, *output_image_b = NULL;
// int kernel_size; double sigma;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 3) {
Erro("Error: Wrong number of parameters. The rotate_nl_filter must have two parameters <x_offset> <y_offset>.", "", "");
return;
}
// Gets the Filter Parameters - The Pointers And The Values - Void pointers must be casted
// kernel_size = filter_desc->filter_params->next->param.ival;
// sigma = filter_desc->filter_params->next->next->param.fval;
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
//wo = nl_output->dimentions.x;
//ho = nl_output->dimentions.y;
// input_image_b[i] = (float)(BLUE(nl_input->neuron_vector[i].output.ival) / 255.0);
// input_image_g[i] = (float)(GREEN(nl_input->neuron_vector[i].output.ival) / 255.0);
// input_image_r[i] = (float)(RED(nl_input->neuron_vector[i].output.ival) / 255.0);
for(i = 0; i < wi * hi; i++)
nl_output->neuron_vector[i].output.ival = PIXEL((long int)(255 * output_image_r[i]), (long int)(255 * output_image_g[i]), (long int)(255 * output_image_b[i]));
}
//---------------------------------------------------------------------------
bool __fastcall TRawPrint::NovaPagina(void)
{
bool Resultado = true;
try
{
if (DispAberto)
{
if (PaginaAberta)
if (!EndPagePrinter(hPrinter))
Erro("Erro ao finalizar a Página");
if (!StartPagePrinter(hPrinter))
Resultado = false;
else
PaginaAberta = true;
}
}
catch(Exception & E)
{
Erro("Erro ao gerar nova Página.\r\n\r\n" + E.Message + "\r\n");
}
return(Resultado);
};
//---------------------------------------------------------------------------
bool __fastcall TRawPrint::ImprimirLinha(AnsiString Linha)
{
DWORD WrittenChars;
bool Retorno = true;
try
{
if (DispAberto)
if (!WritePrinter(hPrinter, Linha.c_str(), Linha.Length(), &WrittenChars))
Retorno = false;
}
catch(Exception & E)
{
Erro("Erro ao Imprimir Linha.\r\n\r\n" + E.Message + "\r\n");
}
return(Retorno);
};
void
init_character_input(INPUT_DESC *input)
{
#ifndef NO_INTERFACE
int x, y;
#endif
float f;
char file_name[256];
strcpy(file_name, input->name);
strcat(file_name, ".in");
if ((g_input_file = fopen(file_name, "r")) == NULL)
Erro("cannot open input file: ", file_name, "");
set_train_data_set(&character);
make_input_image_character(input);
f = 1.0;
while ((((float)input->ww * f) < 128.0) || (((float)input->wh * f) < 128.0))
f += 1.0;
while ((((float)input->ww * f) > 1024.0) || (((float)input->wh * f) > 1024.0))
f *= 0.9;
#ifndef NO_INTERFACE
glutInitWindowSize((int)((float)input->ww * f),(int)((float)input->wh * f));
if (read_window_position(input->name, &x, &y))
glutInitWindowPosition(x, y);
else
glutInitWindowPosition(-1, -1);
input->win = glutCreateWindow(input->name);
glGenTextures(1,(GLuint *)(& (input->tex)));
input_init(input);
glutReshapeFunc(input_reshape);
glutDisplayFunc(input_display);
glutMouseFunc(input_mouse);
glutPassiveMotionFunc(input_passive_motion);
glutKeyboardFunc(keyboard);
#endif
}
void hamming_distance_inhibition_filter(FILTER_DESC *filter_desc)
{
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
int xo, yo, wo, ho;
//float avg_distance;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 1) {
Erro("Error: Wrong number of parameters. The hamming_distance_filter must have no parameters", "", "");
return;
}
// Gets the Filter Output
nl_output = filter_desc->output;
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
// Copies the hamming distance into the output layer
for (yo = 0; yo < ho; yo++)
for (xo = 0; xo < wo; xo++)
nl_output->neuron_vector[xo + yo * wo].output.fval = nl_input->neuron_vector[xo + yo * wo].output.fval * angular_similarity(sqrt(hamming_squared_cosine_int(nl_input->neuron_vector[xo + yo * wo].last_hamming_distance,INPUTS_PER_NEURON)));
}
short compara(char *um, char *dois){
unsigned int lenght;
if(um == NULL || dois == NULL)
Erro("Null string ordenando");
lenght = min(strlen(um), strlen(dois));
do{
if(*um == *dois){
um++;
dois++;
}
if(tolower(*um) < tolower(*dois))
return 1;
if(tolower(*dois) < tolower(*um))
return 2;
}while(--lenght);
return strlen(um) < strlen(dois) ? 1 : 2;
}
void refresh(char *name, char *alvo){
char *buf, *ptr, *ptr2, *tmp;
FILE *target;
if(name == NULL)
return;
if(alvo == NULL)
target = stdout;
else
target = Fopen(alvo, "w");
ptr = buf = pega(name);
while(1){
if(*ptr == '<'){
if(++ptr == '\0')
break;
else
if(ptr == strstr(ptr, "hts")){
ptr2 = ptr + strlen("hts");
ptr = tags(ptr2);
if(ptr++ == NULL)
Erro(">");
tmp = Malloc(ptr - ptr2);
strncpy(tmp, ptr2, ptr - ptr2 - 1);
processa(tmp, target);
}
else fputc('<', target);
}
if(*ptr == '\0') break;
fputc(*(ptr++), target);
}
if(target != stdout)
fclose(target);
free(buf);
}
void
read_data_set_order(int *order, int size, const char *file_name)
{
int i = 0;
FILE *file;
if ((file = fopen(file_name, "r")) == NULL)
{
while(i < size)
{
order[i] = i;
i++;
}
}
else
{
while(!feof(file))
{
if (fscanf(file, "%d\n", &order[i]) < 0)
Erro((char *)"Error reading order from file", (char *)file_name, (char *)"");
i++;
}
}
}
void empilha(double x)
{
if (topo < TAMPIL) pilha[topo++] = x;
else Erro("Pilha cheia");
}
double desempilha()
{
if (topo > 0) return pilha[--topo];
else Erro ("Pilha vazia");
return 1.0;
}
void robot_reflectance_filter (FILTER_DESC *filter_desc)
{
NEURON_LAYER_LIST *n_list;
NEURON_LAYER *nl_image, *nl_illuminance;
PARAM_LIST *p_list;
int i, j;
int w, h;
int pixel;
float I, L, R;
for (i = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, i++);
if (i != 2)
{
Erro ("Wrong number of neuron layers. The robot_reflectance_filter must be applied on two neuron layers.", "", "");
return;
}
if (filter_desc->neuron_layer_list->next->neuron_layer->output_type != GREYSCALE_FLOAT)
{
Erro ("The output type of the illuminance neuron layer must be greyscale_float.", "", "");
return;
}
if (filter_desc->output->output_type != GREYSCALE_FLOAT)
{
Erro ("The output type of the robot_reflectance_filter output must be greyscale_float.", "", "");
return;
}
for (i = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, i++);
i--;
if (i != 0)
{
Erro ("Wrong number of parameters. The robot_reflectance_filter has no parameters.", "", "");
return;
}
nl_image = filter_desc->neuron_layer_list->neuron_layer;
nl_illuminance = filter_desc->neuron_layer_list->next->neuron_layer;
w = filter_desc->output->dimentions.x;
h = filter_desc->output->dimentions.y;
for (i = 0; i < w; i++)
{
for (j = 0; j < h; j++)
{
pixel = nl_image->neuron_vector[j * w + i].output.ival;
I = (float) (RED (pixel) + GREEN (pixel) + BLUE (pixel)) / 3.0;
L = nl_illuminance->neuron_vector[j * w + i].output.fval;
if (I == 0.0)
R = 0.0;
else if (L == 0.0)
R = 1.0;
else
R = I / L;
filter_desc->output->neuron_vector[j * w + i].output.fval = R;
}
}
}
void
hsv_h_filter(FILTER_DESC *filter_desc)
{
// http://en.wikipedia.org/wiki/HSL_and_HSV
// http://cs.haifa.ac.il/hagit/courses/ist/Lectures/Demos/ColorApplet2/t_convert.html
PARAM_LIST *p_list = NULL;
NEURON_LAYER_LIST *n_list = NULL;
NEURON_LAYER *nl_output = NULL, *nl_input = NULL;
int nl_number, p_number;
int xo, yo, wi, hi, wo, ho;
int r, g, b, max, min;
double h, delta;
// Checks the Neuron Layers Number
for (nl_number = 0, n_list = filter_desc->neuron_layer_list; n_list != NULL; n_list = n_list->next, nl_number++)
;
// Checks the Parameters Number
for (p_number = 0, p_list = filter_desc->filter_params; p_list != NULL; p_list = p_list->next, p_number++)
;
if (p_number != 1)
{
Erro ("Error: Wrong number of parameters. The blue_mask_filter must have no parameters.", "", "");
return;
}
// Gets the Input Neuron Layer
nl_input = filter_desc->neuron_layer_list->neuron_layer;
// Gets the Filter Output
nl_output = filter_desc->output;
// Input-Output width
wi = nl_input->dimentions.x;
hi = nl_input->dimentions.y;
wo = nl_output->dimentions.x;
ho = nl_output->dimentions.y;
if (wi != wo || hi != ho)
{
Erro ("Error: Input and output layers on blue_channel_mask filter must have the same 2D size.", "", "");
return;
}
for (yo = 0; yo < ho; yo++)
{
for (xo = 0; xo < wo; xo++)
{
r = RED(nl_input->neuron_vector[xo + yo * wo].output.ival);
g = GREEN(nl_input->neuron_vector[xo + yo * wo].output.ival);
b = BLUE(nl_input->neuron_vector[xo + yo * wo].output.ival);
max = max_value(r, g, b);
min = min_value(r, g, b);
if ((max != 0) && (max != min))
{
delta = (double) max - (double) min;
if (r == max)
h = (double) (g - b) / delta; // between yellow & magenta
else if (g == max)
h = 2.0 + (double) (b - r) / delta; // between cyan & yellow
else
h = 4.0 + (double) (r - g) / delta; // between magenta & cyan
h *= 60.0; // degrees
if (h < 0.0)
h += 360.0;
nl_output->neuron_vector[xo + yo * wo].output.ival = (int) (255.0 * (h / 360.0) + 0.5);
}
else
nl_output->neuron_vector[xo + yo * wo].output.ival = 0;
}
}
}