Box blur implementation

**JPEG23** · 08-04-2020

Having trouble with this - I know I've written way too much code but I can't figure out what different to do...

RGB is a

Code:

struct

containing the red green and blue

Code:

void blur(int height, int width, RGB image[height][width])
{
    for (int i = 0; i < height; i++)
    {
        for (int j = 0; j < width; j++)
        {
                //0,0
                float red1 = image[i][j].rgbtRed;
                float green1 = image[i][j].rgbtGreen;
                float blue1 = image[i][j].rgbtBlue;
                
                //0,1
                float red2 = image[i][j + 1].rgbtRed;
                float green2 = image[i][j + 1].rgbtGreen;
                float blue2 = image[i][j + 1].rgbtBlue;
                
                //0,2
                float red3 = image[i][j + 2].rgbtRed;
                float green3 = image[i][j + 2].rgbtGreen;
                float blue3 = image[i][j + 2].rgbtBlue;
                
                //1,0
                float red4 = image[i + 1][j].rgbtRed;
                float green4 = image[i + 1][j].rgbtGreen;
                float blue4 = image[i + 1][j].rgbtBlue;
                
                //1,1
                float red5 = image[i + 1][j + 1].rgbtRed;
                float green5 = image[i + 1][j + 1].rgbtGreen;
                float blue5 = image[i + 1][j + 1].rgbtBlue;
                
                //1,2
                float red6 = image[i + 1][j + 2].rgbtRed;
                float green6 = image[i + 1][j + 2].rgbtGreen;
                float blue6 = image[i + 1][j + 2].rgbtBlue;
            
                //2,0
                float red7 = image[i + 2][j].rgbtRed;
                float green7 = image[i + 2][j].rgbtGreen;
                float blue7 = image[i + 2][j].rgbtBlue;
                
                //2,1
                float red8 = image[i + 2][j + 1].rgbtRed;
                float green8 = image[i + 2][j + 1].rgbtGreen;
                float blue8 = image[i + 2][j + 1].rgbtBlue;
                
                //2,2
                float red9 = image[i + 2][j + 2].rgbtRed;
                float green9 = image[i + 2][j + 2].rgbtGreen;
                float blue9 = image[i + 2][j + 2].rgbtBlue;
                
                float averageRed = red1 + red2 + red3 + red4 + red5 + red6 + red7 + red8 + red9;
                float averageGreen = green1 + green2 + green3 + green4 + green5 + green6 + green7 + green8 + green9;
                float averageBlue = blue1 + blue2 + blue3 + blue4 + blue5 + blue6 + blue7 + blue8 + blue9;
                
                averageRed = averageRed / 9;
                averageGreen = averageGreen / 9;
                averageBlue = averageBlue / 9;
                
                image[i][j].rgbtRed = roundf(averageRed);
                image[i][j].rgbtGreen = roundf(averageGreen);
                image[i][j].rgbtBlue = roundf(averageBlue);

**Salem** · 08-04-2020

> float red2 = image[i][j + 1].rgbtRed;
All these +1 and +2 are going to overflow your array bounds at the limits of your loops.

Code:

float red = 0;
float green = 0;
float blue = 0;
for ( int x = 0 ; x < 2 ; x++ ) {
  for ( int y = 0 ; y < 2 ; y++ ) {
    red += image[i+x][j+y].rgbtRed;
  }
}

Saves you copy/pasting 40 lines of code.

**JPEG23** · 08-04-2020

Originally Posted by Salem

> float red2 = image[i][j + 1].rgbtRed;
All these +1 and +2 are going to overflow your array bounds at the limits of your loops.

Code:

float red = 0;
float green = 0;
float blue = 0;
for ( int x = 0 ; x < 2 ; x++ ) {
  for ( int y = 0 ; y < 2 ; y++ ) {
    red += image[i+x][j+y].rgbtRed;
  }
}

Saves you copy/pasting 40 lines of code.

Yes you're correct that's exactly what happened when I ran the program.

**Malcolm McLean** · 08-04-2020

Originally Posted by Salem

> float red2 = image[i][j + 1].rgbtRed;
All these +1 and +2 are going to overflow your array bounds at the limits of your loops.

Code:

float red = 0;
float green = 0;
float blue = 0;
for ( int x = 0 ; x < 2 ; x++ ) {
  for ( int y = 0 ; y < 2 ; y++ ) {
    red += image[i+x][j+y].rgbtRed;
  }
}

Saves you copy/pasting 40 lines of code.

This is probably what you want for a box filter. a 3x3 box
centred on the target pixel.
You can be a bit more sophisticated in handling the edges.

You can't write directly to your source as you corrupt the image
for the next pixel.

Code:

float red = 0;
float green = 0;
float blue = 0;
int N = 0;
for ( int x = -1 ; x < 2 ; x++ ) {
  for ( int y = -1 ; y < 2 ; y++ ) {
    if (i + x >= 0 && i + x < width && j + y >= 0 && j + y < height){
      red += image[i+x][j+y].rgbtRed;
      N++;
    }
  }
}
outputimage[j][i].rgbtRed = red / N;

**JPEG23** · 08-04-2020

Originally Posted by Malcolm McLean

This is probably what you want for a box filter. a 3x3 box
centred on the target pixel.
You can be a bit more sophisticated in handling the edges.

You can't write directly to your source as you corrupt the image
for the next pixel.

Code:

float red = 0;
float green = 0;
float blue = 0;
int N = 0;
for ( int x = -1 ; x < 2 ; x++ ) {
  for ( int y = -1 ; y < 2 ; y++ ) {
    if (i + x >= 0 && i + x < width && j + y >= 0 && j + y < height){
      red += image[i+x][j+y].rgbtRed;
      N++;
    }
  }
}
outputimage[j][i].rgbtRed = red / N;

Thanks. I'm not going to lie and pretend that I understand how/where to implement the changes you've offered! What are the underlying concepts behind this box blur implementation in C?
Specifically, what areas of C should I learn more about to deal with this? The problem came alongside teaching about memory, so I assume it's that? I try not to look at other people's code but in the little snippets I have seen, I've seen people using malloc.

At the moment, I'm having fundamental problems understanding what I can and cannot do with the syntax e.g. how do I access the surrounding pixels in order to calculate their averages - hence the +1/-1 type approach I took.

The code I wrote was my naive way of implementing the following:

1) Assign each color within each pixel to a variable (float red1, green1 etc)

2) Work out the averages

3) Assign the averages to the current pixel (i.e. image[i][j])

Thread: Box blur implementation

Thread Tools

Search Thread

Display

Box blur implementation

Similar Threads

Image Blur Filter

Quadratic and Gaussian Blur

Blur and sharpen a bitmap

Window Drag / Blur

Makes Me So Blur!!!

Tags for this Thread