Roxame Home - Diccan Home

Measurement, evaluation and critique

Main commands

All these values can be displayed with affeval . That will display also the number of regions, which matters to evaluate an image.

affcol scrolls the parameters of current color

affdifg scrolls the difg
affeval scrolls evaluation
affhisthue displays an histogram of the hues
affhistsat
affhistval
affhsv
affich
affmood
affnbreg
affnumber
affpas
affreduc assigns affrReduc to number
affreg
affrgb
affstkl
afftdc
afftdifg
affxd
affxorig

Philosophy

Roxame has several measurement algorithms, and uses them in different phases of its process.

At present, these functions apply mainly to images.
For text, there is an measurument of its length.

Note the two elements appreciation in diccan (number of calls, value) and in the base

We have several functions, which could be considered as descrriptive :
- simple measurement functions
-  recognition functions (forms, colors...)
- classification and segmentation,
- a global function of evaluation and critique  

The basic evaluations produce numbers. Critique is a global evaluation producing a text.

The main one, and the most frequently used, is "complexity" (called difg), a rather crude measurement, but efficient : the number of pixels differents from the the upper and left pixels.

 
- difh is the proportion of pixels differing from the upper one
- difv is the proportion of pixels differing from the left one
- dgen is the number of non white pixels
- dnsR, dnsG et dnsB give the percentage of pixels having some quantity of this color; note that, for a white page, these three fitgures are 100
- dsat is the average saturation (ssat)
- dval is the average value (vval) 

We have several functions, which could be considered as descrriptive :
- simple measurement functions
-  recognition functions (forms, colors...)
- classification and segmentation,
- a global function of evaluationand critique  

Critique

As a text, is considered a format : a text commenting a work.
A set of weight criteria. We could consider it as a kind of neuronal network, possibly adaptable to learning

Main parts :

generalSurface (notably difg)
Use_of_color
Use_of_Luminosity
Ranges
Contrast
Regions number of regions
Roxmood Subjective evaluation by Roxame
Subject recognition (works poorly)

Development Hints

The code ; A_measure. Measurement ane evaluations

Main variables 

  int cardreg ;
  int  regmax,  nbregutilx ;
  int xd = 0;  int yd = 0;  int xf= wwidth -1 ;  int yf= wheight - 1 ; 
  int lobpix, hobpix;
  int[][] semclass ; //= new int [wwidth][wheight]; 
  int MaxHue, MinHue, MaxSat, MinSat, MaxVal,MinVal;
int HueDom,  ValHueDom  ;
  float[] minCoul = new float[3]; float[] maxCoul = new float[3];
  int dval,dgen,difh,difv,difg, dsat, dnsR, dnsG,dnsB , totdifcoul
  int  dwrm, dvif ;
  int[] histo = new int[360];
  int totval, tothistob;
  int nbvalhisto, nbvalhistoh, nbvalhistos, nbvalhistov;
  int[][] bhi = new int[360][48];
  int quadType;
int[][] winnerregion = new int[3][2];

 //analyses after face_detect
int cpxGain = 0 ;
int szeGain = 0 ;
int dsRGain = 0 ;
int satGain = 0 ;
int valGain = 0 ;
int focDepx = 0 ;
int focDepy = 0 ;
int skinHue = 0 ; 
int skinVal = 0 ;
String workTypename

Main functions 

M() a series of measurements on the displayed image
Histo : hitograms computation on the displayed image
Averagecolreg : average color of a region
Averagecolclasse : average color of a class