Image processing and sizes

Anticross

is there a difference between pixel loops in different types of images ?

ZapB

What does QImage::format() return for an image that works and one that doesn't?

Anticross

I've got massages in output window of my debbuger
QImage::pixel: coordinate (0,2336) out of range
QImage::pixel: coordinate (0,2337) out of range
QImage::pixel: coordinate (0,2338) out of range
QImage::pixel: coordinate (0,2339) out of range
QImage::pixel: coordinate (0,2340) out of range
QImage::pixel: coordinate (0,2341) out of range
...
So it's happens when I processed 3504*2336 image. why it not limited by size of image or I have a mistake in my loops limits?

ZapB

You have you're i and j loop counters back to front in the call to image.pixel(). It should be:

@
QRgb currentPixel = image.pixel( j, i );
@

since you are using j to loop over the x-coord and i to iterate over the y coord. See how the debugger helps you ;-)

ZapB

You have made the same mistake in the call to setPixel() too.

ZapB

So with all the above in place your code should look like this:

@
void ImageWidget::proccessImage(QColor color)
{
// Search algorithm

QImage image = m_image;
int match = 5;

for(int i = 0; i<image.height(); i++)
{
for(int j = 0; j<image.width(); j++)
{
QRgb currentPixel = image.pixel(j,i);

QColor pixColor = QColor::fromRgb(currentPixel);

if(qAbs(pixColor.red()-color.red()) > match && 
   qAbs(pixColor.green()-color.green()) > match &&
   qAbs(pixColor.blue()-color.blue()) > match)
    image.setPixel(j,i,qRgb(255,255,255));

}
}
m_imageLabel->setPixmap(QPixmap::fromImage(image));
}
@

Anticross

Damn, it's so easy :) I need to be more attentive. Thanks for the help and some code optimization.

giesbert

Hi Anticross,

One idea to make the code a bit more readable und to easier find such problems (use x and y instead of i and j), make some renaming. You can also use QRgb directly.

@
void ImageWidget::proccessImage(QColor color)
{
// Search algorithm

QImage image = m_image;
int difference = 5;

for(int y = 0; y < image.height(); y++)
{
    for(int x = 0; x < image.width(); x++)
    {
        QRgb currentPixel = image.pixel(x,y);

        if( qAbs(qRed(currentPixel) - color.red()) > difference &&
            qAbs(qGreen(currentPixel) - color.green()) > difference &&
            qAbs(qBlue(currentPixel) - color.blue()) > difference)
        {
            image.setPixel(x,y,qRgb(255,255,255));
        }
    }
}
m_imageLabel->setPixmap(QPixmap::fromImage(image));

}
@

ZapB

I usually tend to use i for iterating over x and j for iterating over ysince it reminds me of i and j unit vectors. I also tend to use epsilon or delta for small values such as your match or difference variables. It's just taste though and this is my background in mathematics prejudicing me. ;-)

giesbert

As we are talking about coordinates, and height and width, it's usually x and y. That's why I changed that. i is usually (with the developers I work with :-) ) a variable for a normal for loop. For cascading, I normally use row and column or s9imilar things, por x and y if they are coordinated (which they are in this case).

But it's a matter of taste.

And match was (for me) just nothing that was clear from the name. match for me would be a boolean (matches or not)

ZapB

Yeah for me it also depends upon the context in which the loop(s) occur. If I am iterating over rows then I'll use "row". Usually I try to use something that makes it obvious to which the counter refers.