Save Image into a file in C++ - c++

Question

I want to save an image into a custom file, in order to secure all images of my program.
I tried to save every pixels (as Uint32) of an image like this (I am using SFML)
void fromPNGtoCustomFile(sf::Texture texture, std::string path)
{
std::ofstream fo;
fo.open(path);
sf::Image image=texture.copyToImage(); //GET IMAGE OF THE TEXTURE
fo << image.getSize().x << " " << image.getSize().y << " "; // WRITE THE SIZE OF THE IMAGE
for(unsigned int i=0; i< image.getSize().x; i++)
{
for(unsigned int j=0; j< image.getSize().y; j++)
{
fo << image.getPixel(i, j).toInteger() << " ";
// image.getPixel(x, y).toInteger() RETURNS A Uint32 VALUE
}
}
fo.close();
}
Then I load image from the file using the same logic.
It worked, but I realised that the size of the file I created was around 250 Mb, when my original .png image was only 8 Mb. If I compress the new file using 7Zip, I get a 8 Mb file, as the original size. I do not understand why I get this size.
I don't really know what is the best way to create custom file for saving images.
Do you have any suggestions, or correction of my code?

Answer 1

To answer the question of why you are getting such a big file size, it's because you are writing the information for every single pixel independently, so your file ends up as large as it can be. In order to reduce the size, the compressed formats take advantage of, for instance, the fact that there is usually a large number of neighbouring pixels that are the same colour. For example, an image of the sky will very likely contain an immense amount of pixels of the same shade of blue. So what you need to do is to define an algorithm with some sort of special code to denote that "the next N pixels are some RGB colour" or something to that effect.
A good source on data compression for images is on this chapter of the DSP Guide: http://www.dspguide.com/ch27.htm