Subject:
|
colour sensor caliberation
|
Newsgroups:
|
lugnet.robotics.handyboard
|
Date:
|
Sat, 26 Feb 2000 03:55:47 GMT
|
Original-From:
|
E158 (M.FAZAL) <m.fazal@bham-eee-fsSPAMCAKE4.bham.ac.uk>
|
Viewed:
|
1124 times
|
| |
|
|
Hi all
I mailed the list a few weeks back about implementing a colour sensor.
someone suggested using three photo resistors with red, green and blue
filters over it.
I have built the sensor with four photo resitors, three with the different
filters and one without the filter. i have also written a caliberating
software and a program to use the sensor.Anyway, the result is that
sometimes the sensor gives the correct colour. But gets it wrong quite often
too. usually it confuses green for blue or brown (and other way round). also
if it is too closer than the calibertion distance it thinks the surface is
black. below is the method i am using.
struct surface
{
char colour[5];
int no_filter; /* value from photoresistor with no filter */
int red_filter; /* value from photoresistor with red filter */
int blue_filter; /* value from photoresistor with blue filter */
int green_filter; /* value from photoresistor with green filter */
float red_ratio; /* red filter value / no filter */
float blue_ratio; /* blue filter value / no filter */
float green_ratio; /* green filter value / no filter */
};
persistent struct surface s1,s2,s3,s4,s5,s6; /* 6 surfaces with different
colours */
struct surface* surf_no; /* pointer to surface */
.
.
void cal_values() /* function to caliberate the values */
{
printf("Caliberating. Please wait...\n");
surf_no->no_filter=analog(2); /* \ */
surf_no->red_filter=analog(3); /* \ values from analog ports */
surf_no->blue_filter=analog(0); /* / */
surf_no->green_filter=analog(1); /* / */
surf_no->red_ratio=(float)surf_no->red_filter/float)surf_no->no_filter;
surf_no->blue_ratio=(float)surf_no->blue_filter/float)surf_no->no_filter;
surf_no->green_ratio=(float)surf_no->green_filter/loat)surf_no->no_filter;
}
To determine the surface coulor after caliberation, the sensor is placed in
from of a surface and the values are read, and compared with the values for
each of the 6 surfaces in the caliberation. the colour with the closet value
is the right colour. below is a pice of the code that does this.
void get_colour(void)
{
float err, prev_err=100.0;
no_filter=analog(2);
red_filter=analog(3);
blue_filter=analog(0);
green_filter=analog(1);
red_ratio = (float)red_filter / (float)no_filter;
blue_ratio = (float)blue_filter / (float)no_filter;
green_ratio = (float)green_filter / (float)no_filter;
surf_no=&s1;
s=surf_no; /* s is a pointer to surface which is the most likely
colour */
err=compare();
if(err<prev_err){
s=surf_no;
prev_err=err;
}
. /* do same comparing for all other caliberated colours */
.
printf("Surface colour is %s \n",s->colour);
}
float compare()
{
float tot_err, red_err, blue_err, green_err;
red_err = surf_no->red_ratio - red_ratio; /* caliberated value - current
surface value */
if (red_err<0.0) red_err = 0.0 - red_err; /* make value positive */
blue_err = surf_no->blue_ratio - blue_ratio; /* caliberated value - current
surface value */
if (blue_err<0.0) blue_err = 0.0 - blue_err; /* make value positive */
green_err = surf_no->green_ratio - green_ratio;
if (green_err<0.0) green_err= 0.0 - green_err; /* make value positive */
tot_err = red_err + blue_err + green_err; /* sum of all errors */
return tot_err; /* return error value */
}
I would be most grateful if you have any suggetsions about my method for
caliberating and correlating the values. Any ideas most welcome.
Many thanks
M Fazal
|
|
1 Message in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
