vq/metrics.c

   1 /********************************************************************
   2  *                                                                  *
   3  * THIS FILE IS PART OF THE Ogg Vorbis SOFTWARE CODEC SOURCE CODE.  *
   4  * USE, DISTRIBUTION AND REPRODUCTION OF THIS SOURCE IS GOVERNED BY *
   5  * THE GNU PUBLIC LICENSE 2, WHICH IS INCLUDED WITH THIS SOURCE.    *
   6  * PLEASE READ THESE TERMS DISTRIBUTING.                            *
   7  *                                                                  *
   8  * THE OggSQUISH SOURCE CODE IS (C) COPYRIGHT 1994-2000             *
   9  * by Monty <monty@xiph.org> and The XIPHOPHORUS Company            *
  10  * http://www.xiph.org/                                             *
  11  *                                                                  *
  12  ********************************************************************
  13
  14  function: function calls to collect codebook metrics
  15  last mod: $Id: metrics.c,v 1.1 2000/01/06 13:57:13 xiphmont Exp $
  16
  17  ********************************************************************/
  18
  19
  20 #include <stdlib.h>
  21 #include <unistd.h>
  22 #include <math.h>
  23 #include "bookutil.h"
  24
  25 /* set up metrics */
  26
  27 double meanamplitude_acc=0.;
  28 double meanamplitudesq_acc=0.;
  29 double meanerror_acc=0.;
  30 double meanerrorsq_acc=0.;
  31 double meandev_acc=0.;
  32
  33 double *histogram=NULL;
  34 double *histogram_errorsq=NULL;
  35 double *histogram_distance=NULL;
  36 double *histogram_hi=NULL;
  37 double *histogram_lo=NULL;
  38
  39 double count=0.;
  40
  41 int histerrsort(const void *a, const void *b){
  42   double av=histogram_distance[*((long *)a)];
  43   double bv=histogram_distance[*((long *)b)];
  44   if(av<bv)return(-1);
  45   return(1);
  46 }
  47
  48 void process_preprocess(codebook *b,char *basename){
  49   histogram=calloc(b->entries,sizeof(double));
  50   histogram_distance=calloc(b->entries,sizeof(double));
  51   histogram_errorsq=calloc(b->entries*b->dim,sizeof(double));
  52   histogram_hi=calloc(b->entries*b->dim,sizeof(double));
  53   histogram_lo=calloc(b->entries*b->dim,sizeof(double));
  54 }
  55
  56 void process_postprocess(codebook *b,char *basename){
  57   int i,j,k;
  58   char *buffer=alloca(strlen(basename)+80);
  59
  60   fprintf(stderr,"Done.  Processed %ld data points for %ld entries:\n",
  61           (long)count,b->entries);
  62   fprintf(stderr,"\tglobal mean amplitude: %g\n",
  63           meanamplitude_acc/(count*b->dim));
  64   fprintf(stderr,"\tglobal mean squared amplitude: %g\n",
  65           sqrt(meanamplitudesq_acc/(count*b->dim)));
  66
  67   fprintf(stderr,"\tglobal mean error: %g\n",
  68           meanerror_acc/(count*b->dim));
  69   fprintf(stderr,"\tglobal mean squared error: %g\n",
  70           sqrt(meanerrorsq_acc/(count*b->dim)));
  71   fprintf(stderr,"\tglobal mean deviation: %g\n",
  72           meandev_acc/(count*b->dim));
  73   {
  74     FILE *out;
  75     sprintf(buffer,"%s-fit.m",basename);
  76     out=fopen(buffer,"w");
  77     if(!out){
  78       fprintf(stderr,"Could not open file %s for writing\n",buffer);
  79       exit(1);
  80     }
  81
  82     for(i=0;i<b->entries;i++){
  83       for(k=0;k<b->dim;k++){
  84         fprintf(out,"%d, %g, %g\n",
  85                 i*b->dim+k,(b->valuelist+i*b->dim)[k],
  86                 sqrt((histogram_errorsq+i*b->dim)[k]/histogram[i]));
  87       }
  88     }
  89     fclose(out);
  90
  91     sprintf(buffer,"%s-worst.m",basename);
  92     out=fopen(buffer,"w");
  93     if(!out){
  94       fprintf(stderr,"Could not open file %s for writing\n",buffer);
  95       exit(1);
  96     }
  97
  98     for(i=0;i<b->entries;i++){
  99       for(k=0;k<b->dim;k++){
 100         fprintf(out,"%d, %g, %g, %g\n",
 101                 i*b->dim+k,(b->valuelist+i*b->dim)[k],
 102                 (b->valuelist+i*b->dim)[k]+(histogram_lo+i*b->dim)[k],
 103                 (b->valuelist+i*b->dim)[k]+(histogram_hi+i*b->dim)[k]);
 104       }
 105     }
 106     fclose(out);
 107   }
 108
 109   {
 110     FILE *out;
 111     long *index=alloca(sizeof(long)*b->entries);
 112     sprintf(buffer,"%s-distance.m",basename);
 113     out=fopen(buffer,"w");
 114     if(!out){
 115       fprintf(stderr,"Could not open file %s for writing\n",buffer);
 116       exit(1);
 117     }
 118     for(j=0;j<b->entries;j++){
 119       if(histogram[j])histogram_distance[j]/=histogram[j];
 120       index[j]=j;
 121     }
 122
 123     qsort(index,b->entries,sizeof(long),histerrsort);
 124
 125     for(j=0;j<b->entries;j++)
 126       for(k=0;k<histogram[index[j]];k++)
 127         fprintf(out,"%g,\n",histogram_distance[index[j]]);
 128     fclose(out);
 129
 130   }
 131
 132 }
 133
 134 void process_vector(codebook *b,double *a){
 135   int entry=codebook_entry(b,a);
 136   double *e=b->valuelist+b->dim*entry;
 137   int i;
 138   double amplitude=0.;
 139   double distance=0.;
 140   double base=0.;
 141   for(i=0;i<b->dim;i++){
 142     double error=a[i]-e[i];
 143     if(b->q_sequencep){
 144       amplitude=a[i]-base;
 145       base=a[i];
 146     }else
 147       amplitude=a[i];
 148
 149     meanamplitude_acc+=fabs(amplitude);
 150     meanamplitudesq_acc+=amplitude*amplitude;
 151     meanerror_acc+=fabs(error);
 152     meanerrorsq_acc+=error*error;
 153
 154     if(amplitude)
 155       meandev_acc+=fabs((a[i]-e[i])/amplitude);
 156     else
 157       meandev_acc+=fabs(a[i]-e[i]); /* yeah, yeah */
 158
 159     histogram_errorsq[entry*b->dim+i]+=error*error;
 160     if(histogram[entry]==0 || histogram_hi[entry*b->dim+i]<error)
 161       histogram_hi[entry*b->dim+i]=error;
 162     if(histogram[entry]==0 || histogram_lo[entry*b->dim+i]>error)
 163       histogram_lo[entry*b->dim+i]=error;
 164     distance+=error*error;
 165   }
 166
 167   histogram[entry]++;
 168   histogram_distance[entry]+=sqrt(distance);
 169
 170   if((long)(count++)%100)spinnit("working.... lines: ",count);
 171 }
 172
 173 void process_usage(void){
 174   fprintf(stderr,
 175           "usage: vqmetrics <codebook>.vqh datafile.vqd [datafile.vqd]...\n\n"
 176           "       data can be taken on stdin.  Output goes to output files:\n"
 177           "       basename-fit.m:      gnuplot: mean square error by entry value\n"
 178           "       basename-worst.m:    gnuplot: worst error by entry value\n"
 179           "       basename-distance.m: gnuplot file showing distance probability\n"
 180           "\n");
 181
 182 }