/*********************************************************************/

/* dct.c Function to perform a 8 point 2D DCT						 */

/* DCT is performed using direct matrix multiplication				 */

/*********************************************************************/



/* externals */

#include "dct_main.h"



extern unsigned char image_in[IMAGE_SIZE];

extern unsigned char image_out[IMAGE_SIZE];

extern short block[BLOCK_SIZE];

extern const short coe[12];





void dct(void)

{

	short ADD[20];    /* Table of the addition coefficients */

	int M[12];      /* Table of the results of the multiplication */

	int postadd1,postadd2;

	int i,j;



	for(j=0;j<8;j++)

		{

		/* first set of addtions */

		ADD[0]= (block[j]+block[56+j]);   	/* x(0)+x(7) */

		ADD[1]= (block[24+j]+block[32+j]); 	/* x(3)+x(4) */

		ADD[2]= (block[8+j]+block[48+j]);  	/* x(1)+x(6) */

		ADD[3]= (block[16+j]+block[40+j]);  /* x(2)+x(5) */

		ADD[4]= (block[j]-block[56+j]);  	/* x(0)+x(7) */

		ADD[5]= (block[48+j]-block[8+j]);  	/* x(6)-x(1) */

		ADD[6]= (block[24+j]-block[32+j]);  /* x(3)-x(4) */

		ADD[7]= (block[16+j]-block[40+j]);  /* x(2)-x(5) */



		/* second set of addtions, this is done so previous additions

			do not need to be repeated */

		ADD[8]= (ADD[0]+ADD[1]);

		ADD[9]= (ADD[0]-ADD[1]);

		ADD[10]=(ADD[2]+ADD[3]);

		ADD[11]=(ADD[2]-ADD[3]);

		ADD[12]=(ADD[4]+ADD[6]);

		ADD[13]=(ADD[5]+ADD[7]);

		ADD[14]=(ADD[9]+ADD[11]);

		ADD[15]=(ADD[4]+ADD[5]);

		ADD[16]=(ADD[12]+ADD[13]);

		ADD[17]=(ADD[6]+ADD[7]);

		ADD[18]=(ADD[8]+ADD[10]);

		ADD[19]=(ADD[8]-ADD[10]);



		/* Multiplications carried out, note: here 14. Includes

			one over root eight term */

		M[0]= (int)(coe[0]*ADD[9]);

		M[1]= (int)(coe[1]*ADD[14]);

		M[2]= (int)(coe[2]*ADD[11]);

		M[3]= (int)(coe[3]*ADD[4]);

		M[4]= (int)(coe[4]*ADD[15]);

		M[5]= (int)(coe[5]*ADD[5]);

		M[6]= (int)(coe[6]*ADD[12]);

		M[7]= (int)(coe[7]*ADD[16]);

		M[8]=(int)(coe[8]*ADD[13]);

		M[9]=(int)(coe[9]*ADD[6]);

		M[10]=(int)(coe[10]*ADD[17]);

		M[11]=(int)(coe[11]*ADD[7]);



		/* post multiplication, additions + subtractions */

		block[j]=ADD[18];                      /* y(0) */

		block[32+j]=ADD[19];                   /* y(4) */

		block[16+j]=(short)((M[0]+M[1])>>12);  /* y(2) */

		block[48+j]=(short)((M[1]-M[2])>>12);  /* y(6) */

		postadd1= M[6]+M[7];

		postadd2= M[7]+M[8];

		block[56+j]= (short)((M[3]+M[4]+postadd1)>>12);    /* y(7) */

		block[40+j]= (short)((M[4]+M[5]+postadd2)>>12);    /* y(5) */

		block[8+j]= (short)((M[9]+M[10]-postadd1)>>12);   /* y(1) */

		block[24+j]= (short)((postadd2-M[10]-M[11])>>12);  /* y(3) */

		}



	for(i=0;i<64;i+=8)

		{

		/* first set of addtions */

		ADD[0]=(block[i]+block[i+7]);   /* x(0)+x(7) */

		ADD[1]=(block[i+3]+block[i+4]); /* x(3)+x(4) */

		ADD[2]=(block[i+1]+block[i+6]); /* x(1)+x(6) */

		ADD[3]=(block[i+2]+block[i+5]); /* x(2)+x(5) */

		ADD[4]=(block[i]-block[i+7]);	  /* x(0)-x(7) */

		ADD[5]=(block[i+6]-block[i+1]); /* x(6)-x(1) */

		ADD[6]=(block[i+3]-block[i+4]); /* x(3)-x(4) */

		ADD[7]=(block[i+2]-block[i+5]); /* x(2)-x(5) */



		/* second set of addtions, this is done so previous additions

			do not need to be repeated */

		ADD[8]=(ADD[0]+ADD[1]);

		ADD[9]=(ADD[0]-ADD[1]);

		ADD[10]=(ADD[2]+ADD[3]);

		ADD[11]=(ADD[2]-ADD[3]);

		ADD[12]=(ADD[4]+ADD[6]);

		ADD[13]=(ADD[5]+ADD[7]);

		ADD[14]=(ADD[9]+ADD[11]);

		ADD[15]=(ADD[4]+ADD[5]);

		ADD[16]=(ADD[12]+ADD[13]);

		ADD[17]=(ADD[6]+ADD[7]);

		ADD[18]=(ADD[8]+ADD[10]);

		ADD[19]=(ADD[8]-ADD[10]);



		/* Multiplications carried out, note: here 14. Includes

			one over root eight term */

		M[0]= (int)(coe[0]*ADD[9]);

		M[1]= (int)(coe[1]*ADD[14]);

		M[2]= (int)(coe[2]*ADD[11]);

		M[3]= (int)(coe[3]*ADD[4]);

		M[4]= (int)(coe[4]*ADD[15]);

		M[5]= (int)(coe[5]*ADD[5]);

		M[6]= (int)(coe[6]*ADD[12]);

		M[7]= (int)(coe[7]*ADD[16]);

		M[8]=(int)(coe[8]*ADD[13]);

		M[9]=(int)(coe[9]*ADD[6]);

		M[10]=(int)(coe[10]*ADD[17]);

		M[11]=(int)(coe[11]*ADD[7]);



		/* post multiplication, additions + subtractions */

		block[i]=(short)(ADD[18]>>3);  	         /* y(0) */

		block[i+4]=(short)(ADD[19]>>3);           /* y(4) */

		block[i+2]=(short)((M[0]+M[1])>>15);      /* y(2) */

		block[i+6]=(short)((M[1]-M[2])>>15);      /* y(6) */

		postadd1= M[6]+M[7];

		postadd2= M[7]+M[8];

		block[i+7]= (short)((M[3]+M[4]+postadd1)>>15);    /* y(7) */

		block[i+5]= (short)((M[4]+M[5]+postadd2)>>15);    /* y(5) */

		block[i+1]= (short)((M[9]+M[10]-postadd1)>>15);   /* y(1) */

		block[i+3]= (short)((postadd2-M[10]-M[11])>>15);  /* y(3) */

		}



	

}