I am making a memory matching game using playing cards. I want to post the same card twice randomly to the screen.

int a[20] = { 0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9 };

void init()
{
	srand(time(NULL));
	random_shuffle(a, a + 20);
	cout << a[0] << " " << a[1] << " " << a[2] << " " << a[3] << " " << a[4] << " " << a[5] << " " << a[6] << " " << a[7] << endl;
	texture[a[0]] = loadTex("C:\\Users\\Owner\\Desktop\\2_of_clubs.png");
	texture[a[1]] = loadTex("C:\\Users\\Owner\\Desktop\\3_of_clubs.png");
	texture[a[2]] = loadTex("C:\\Users\\Owner\\Desktop\\4_of_clubs.png");
	texture[a[3]] = loadTex("C:\\Users\\Owner\\Desktop\\5_of_clubs.png");
	texture[a[4]] = loadTex("C:\\Users\\Owner\\Desktop\\6_of_clubs.png");
	texture[a[5]] = loadTex("C:\\Users\\Owner\\Desktop\\7_of_clubs.png");
	texture[a[6]] = loadTex("C:\\Users\\Owner\\Desktop\\8_of_clubs.png");
	texture[a[7]] = loadTex("C:\\Users\\Owner\\Desktop\\9_of_clubs.png");
	texture[a[8]] = loadTex("C:\\Users\\Owner\\Desktop\\10_of_clubs.png");
	texture[a[9]] = loadTex("C:\\Users\\Owner\\Desktop\\jack_of_clubs2.png");
	texture[a[10]] = loadTex("C:\\Users\\Owner\\Desktop\\queen_of_clubs2.png");
	texture[a[11]] = loadTex("C:\\Users\\Owner\\Desktop\\king_of_clubs2.png");
	texture[a[12]] = loadTex("C:\\Users\\Owner\\Desktop\\ace_of_clubs.png");
	texture[a[13]] = loadTex("C:\\Users\\Owner\\Desktop\\2_of_diamonds.png");
	texture[a[14]] = loadTex("C:\\Users\\Owner\\Desktop\\3_of_diamonds.png");
	texture[a[15]] = loadTex("C:\\Users\\Owner\\Desktop\\4_of_diamonds.png");
	texture[a[16]] = loadTex("C:\\Users\\Owner\\Desktop\\5_of_diamonds.png");
	texture[a[17]] = loadTex("C:\\Users\\Owner\\Desktop\\6_of_diamonds.png");
	texture[a[18]] = loadTex("C:\\Users\\Owner\\Desktop\\7_of_diamonds.png");
	texture[a[19]] = loadTex("C:\\Users\\Owner\\Desktop\\8_of_diamonds.png");
//	texture[8] = loadTex("C:\\Users\\Owner\\Desktop\\cardBack_blue1.png");
}

I think it would be simpler if you separated the handling and storage of textures from whatever your deck/screen looks like.

May I suggest that you sick with one atlas for all the cards, or programmatically compose them, so you avoid repeating yourself too much.
Alternatively, you could do a bit of both, and have an atlas for the suites, and then add the values programatically.
Alternatively, you could stick with an image for each card, and solve the loading programmatically.

#include <algorithm> // swap
#include <stdint.h>

namespace PRN
{
	inline uint32_t randI (uint32_t time)
	{
		time *= 1664525;
		time ^= (time << 16);
		time ^= (time >> 16);

		time *= 16807;
		time ^= (time << 8);
		time ^= (time >> 24);

		time *= 100069;
		time ^= (time << 24);
		time ^= (time >> 8);

		time *= 1000099;
		
		return time;
	}
	
	
	template <typename T>
	inline void Shuffle (T *arr, const int n, const int seed = 0)
	{
		int32_t s = seed;
		for (int i=n; i-->0;)
		{
			int j = randI (s++) % (i + 1);
			std::swap(arr[i], arr[j]);
		}
	}
};

int a[20] = { 0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9 };

const int iterations = 5; // one would do, but randomness is increased if doing more often
for (int i=0; i<iterations; i++)
	PRN::Shuffle (a, 20, i*20);
// now the values in a are in (pseudo) random order

JoeJ said:

Obnoxious shuffle code

"randomness is increased if doing more often" is horrible advice IMO.
Use the existing std::shuffle - it's much simpler.

There is a std::shuffle?
ugh… if only i knew this before :D

SuperVGA said:
"randomness is increased if doing more often" is horrible advice IMO.

You may be right. But i use just a hash function over ordered sequence, so it isn't that great after just one run.