interpolate8x8.c

用MPEG-4对YUV视频文件编码压缩成divx视频文件

	      	dest3 = FUNSHIFT2(src5,src4);   
	      		
	      	*(ptr_dst)  = QUADAVG(dest0,dest2);
	      	*(ptr_dst+1)= QUADAVG(dest1,dest3); 
	      	
	      	ptr_src += stride1;
	      	ptr_dst += stride1;
	       }
	}
	 else if(temp==3)
	{
		for (j = 0; j < 8; j++) {
	      	src0 = *(ptr_src);
	      	src1 = *(ptr_src+1);
	      	src2 = *(ptr_src+2);
	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	
	      	dest0 = FUNSHIFT1(src1,src0);   
	      	dest1 = FUNSHIFT1(src2,src1);   
	      	dest2 = FUNSHIFT1(src4,src3);   
	      	dest3 = FUNSHIFT1(src5,src4);   
	      		
	      	*(ptr_dst)  = QUADAVG(dest0,dest2);
	      	*(ptr_dst+1)= QUADAVG(dest1,dest3); 
	      	
	      	ptr_src += stride1;
	      	ptr_dst += stride1;
	       }
	}
}

#else

void
interpolate8x8_halfpel_v_c(uint8_t * const dst,/* vertical half pixel interpolate image */
						   const uint8_t * const src,/* integer pixel image */
						   const uint32_t stride,/* image stride */
						   const uint32_t rounding)/* rounding bit(0 ro 1) */
{
	uint32_t j;
	uint32_t src0,src1,src2,src3,src4,src5,dest0,dest1,dest2,dest3;  
	uint32_t * ptr_dst = (uint32_t *)dst;
	const uint32_t * ptr_src = (uint32_t *)src;
	uint32_t stride1=stride>>2;
	uint32_t temp;
	temp = ((uint32_t)src)&3;
	
	if(!temp)
	{
       	  src0 = *(ptr_src);
      	  src1 = *(ptr_src+1);
	      for (j = 0; j < 8; j++) {
	      	src2 = *(ptr_src+stride1);
	      	src3 = *(ptr_src+stride1+1);
	      	
	      	*(ptr_dst)  = QUADAVG(src0,src2);
	      	*(ptr_dst+1)= QUADAVG(src1,src3); 
	      	
	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			src0=src2;
			src1=src3;
	       }
	 }
	 else if(temp==1)
	{
      	src0 = *(ptr_src);
      	src1 = *(ptr_src+1);
      	src2 = *(ptr_src+2);
      	dest0 = FUNSHIFT3(src1,src0);   
      	dest1 = FUNSHIFT3(src2,src1);   
		for (j = 0; j < 8; j++) {
	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	
	      	dest2 = FUNSHIFT3(src4,src3);   
	      	dest3 = FUNSHIFT3(src5,src4);   
	      		
	      	*(ptr_dst)  = QUADAVG(dest0,dest2);
	      	*(ptr_dst+1)= QUADAVG(dest1,dest3); 
	      	
	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			dest0=dest2;
			dest1=dest3;
	       }
	}
	 else if(temp==2)
	{
      	src0 = *(ptr_src);
      	src1 = *(ptr_src+1);
      	src2 = *(ptr_src+2);
      	dest0 = FUNSHIFT2(src1,src0);   
      	dest1 = FUNSHIFT2(src2,src1);   

		for (j = 0; j < 8; j++) {
	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	
	      	dest2 = FUNSHIFT2(src4,src3);   
	      	dest3 = FUNSHIFT2(src5,src4);   
	      		
	      	*(ptr_dst)  = QUADAVG(dest0,dest2);
	      	*(ptr_dst+1)= QUADAVG(dest1,dest3); 
	      	
	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			dest0=dest2;
			dest1=dest3;
	       }
	}
	 else if(temp==3)
	{
      	src0 = *(ptr_src);
      	src1 = *(ptr_src+1);
      	src2 = *(ptr_src+2);
      	dest0 = FUNSHIFT1(src1,src0);   
      	dest1 = FUNSHIFT1(src2,src1);   

		for (j = 0; j < 8; j++) {
	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	
	      	dest2 = FUNSHIFT1(src4,src3);   
	      	dest3 = FUNSHIFT1(src5,src4);   
	      		
	      	*(ptr_dst)  = QUADAVG(dest0,dest2);
	      	*(ptr_dst+1)= QUADAVG(dest1,dest3); 
	      	
	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			dest0=dest2;
			dest1=dest3;
	       }
	}
}
#endif
#endif
/*
   horizontal and vertical half pixel interpolate of 8x8 block
   Add by fyh 2002.12.26
*/
#ifndef _TRIMEDIA
void
interpolate8x8_halfpel_hv_c(uint8_t * const dst,/* horizontal and vertical half pixel interpolate image */
							const uint8_t * const src,/* integer pixel image */
							const uint32_t stride,/* image stride */
							const uint32_t rounding)/* rounding bit(0 ro 1) */
{
	uint32_t i, j;

	for (j = 0; j < 8; j++) {
		for (i = 0; i < 8; i++) {
			int16_t tot =
				src[j * stride + i] + src[j * stride + i + 1] +
				src[j * stride + i + stride] + src[j * stride + i + stride +
												   1];
			tot = ((tot + 2 - rounding) >> 2);
			dst[j * stride + i] = (uint8_t) tot;
		}
	}
}

#else

void
interpolate8x8_halfpel_hv_c(uint8_t * const dst,/* horizontal and vertical half pixel interpolate image */
							const uint8_t * const src,/* integer pixel image */
							const uint32_t stride,/* image stride */
							const uint32_t rounding)/* rounding bit(0 ro 1) */
{
	uint32_t j;
	uint32_t src0,src1,src2,src3,src4,src5;
	uint32_t dest0,dest1,dest2,dest3,dest4,dest5,dest6,dest7,dest8,dest9,
		     dest10;  

	uint32_t * ptr_dst = (uint32_t *)dst;
	const uint32_t * ptr_src = (uint32_t *)src;

	uint32_t stride1=stride>>2;
	uint32_t temp;
	temp = ((uint32_t)src)&3;
	
	if(!temp)
	{
      	  src0 = *(ptr_src);
	      src1 = *(ptr_src+1);
		  src2 = *(ptr_src+2);
	      for (j = 0; j < 8; j++) {

	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	
			/* vertical interpolate */
			dest0= QUADAVG(src0,src3);
	      	dest1= QUADAVG(src1,src4); 
			dest2= QUADAVG(src2,src5);
	      	
            /* horizontal interpolate */
			dest3 = FUNSHIFT3(dest1,dest0);
			dest4 = FUNSHIFT3(dest2,dest1);

			*(ptr_dst)  =QUADAVG(dest0,dest3);
			*(ptr_dst+1)=QUADAVG(dest1,dest4);

	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			
			src0=src3;
			src1=src4;
			src2=src5;
	       }
	 }
	 else if(temp==1)
	{
      	  src0 = *(ptr_src);
	      src1 = *(ptr_src+1);
		  src2 = *(ptr_src+2);
		  dest0 = FUNSHIFT3(src1,src0);   
      	  dest1 = FUNSHIFT3(src2,src1);
          dest2 = UBYTESEL(src2,1);		  
	      for (j = 0; j < 8; j++) {

	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	dest3 = FUNSHIFT3(src4,src3);   
	      	dest4 = FUNSHIFT3(src5,src4); 
			dest5 = UBYTESEL(src5,1);
			
			/* vertical interpolate */
			dest6= QUADAVG(dest0,dest3);
	      	dest7= QUADAVG(dest1,dest4); 
			dest8= QUADAVG(dest2,dest5);
	      	
            /* horizontal interpolate */
			dest9  = FUNSHIFT3(dest7,dest6);
			dest10 = FUNSHIFT3(dest8,dest7);

			*(ptr_dst)  =QUADAVG(dest6,dest9);
			*(ptr_dst+1)=QUADAVG(dest7,dest10);

	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			
			dest0=dest3;
			dest1=dest4;
			dest2=dest5;
	       }
	}
	 else if(temp==2)
	{
      	  src0 = *(ptr_src);
	      src1 = *(ptr_src+1);
		  src2 = *(ptr_src+2);
		  dest0 = FUNSHIFT2(src1,src0);   
      	  dest1 = FUNSHIFT2(src2,src1);
          dest2 = UBYTESEL(src2,2);		  
	      for (j = 0; j < 8; j++) {

	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	dest3 = FUNSHIFT2(src4,src3);   
	      	dest4 = FUNSHIFT2(src5,src4); 
			dest5 = UBYTESEL(src5,2);
			
			/* vertical interpolate */
			dest6= QUADAVG(dest0,dest3);
	      	dest7= QUADAVG(dest1,dest4); 
			dest8= QUADAVG(dest2,dest5);
	      	
            /* horizontal interpolate */
			dest9  = FUNSHIFT3(dest7,dest6);
			dest10 = FUNSHIFT3(dest8,dest7);

			*(ptr_dst)  =QUADAVG(dest6,dest9);
			*(ptr_dst+1)=QUADAVG(dest7,dest10);

	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			
			dest0=dest3;
			dest1=dest4;
			dest2=dest5;
	       }
	}
	 else if(temp==3)
	{
      	  src0 = *(ptr_src);
	      src1 = *(ptr_src+1);
		  src2 = *(ptr_src+2);
		  dest0 = FUNSHIFT1(src1,src0);   
      	  dest1 = FUNSHIFT1(src2,src1);
          dest2 = UBYTESEL(src2,3);		  
	      for (j = 0; j < 8; j++) {

	      	src3 = *(ptr_src+stride1);
	      	src4 = *(ptr_src+stride1+1);
	      	src5 = *(ptr_src+stride1+2);
	      	dest3 = FUNSHIFT1(src4,src3);   
	      	dest4 = FUNSHIFT1(src5,src4); 
			dest5 = UBYTESEL(src5,3);
			
			/* vertical interpolate */
			dest6= QUADAVG(dest0,dest3);
	      	dest7= QUADAVG(dest1,dest4); 
			dest8= QUADAVG(dest2,dest5);
	      	
            /* horizontal interpolate */
			dest9  = FUNSHIFT3(dest7,dest6);
			dest10 = FUNSHIFT3(dest8,dest7);

			*(ptr_dst)  =QUADAVG(dest6,dest9);
			*(ptr_dst+1)=QUADAVG(dest7,dest10);

	      	ptr_src += stride1;
	      	ptr_dst += stride1;
			
			dest0=dest3;
			dest1=dest4;
			dest2=dest5;
	       }
	}

}
#endif