📄 mv-search.c
字号:
m[11] = d[ 8] - d[11];
m[12] = d[12] + d[15];
m[13] = d[13] + d[14];
m[14] = d[13] - d[14];
m[15] = d[12] - d[15];
d[ 0] = m[ 0] + m[ 1];
d[ 1] = m[ 0] - m[ 1];
d[ 2] = m[ 2] + m[ 3];
d[ 3] = m[ 3] - m[ 2];
d[ 4] = m[ 4] + m[ 5];
d[ 5] = m[ 4] - m[ 5];
d[ 6] = m[ 6] + m[ 7];
d[ 7] = m[ 7] - m[ 6];
d[ 8] = m[ 8] + m[ 9];
d[ 9] = m[ 8] - m[ 9];
d[10] = m[10] + m[11];
d[11] = m[11] - m[10];
d[12] = m[12] + m[13];
d[13] = m[12] - m[13];
d[14] = m[14] + m[15];
d[15] = m[15] - m[14];
/*===== sum up =====*/
for (dd=diff[k=0]; k<16; dd=diff[++k])
{
satd += (dd < 0 ? -dd : dd);
}
satd >>= 1;
}
else
{
/*===== sum up =====*/
for (k = 0; k < 16; k++)
{
satd += byte_abs [diff [k]];
}
}
return satd;
}
/*!
***********************************************************************
* \brief
* Sub pixel block motion search
***********************************************************************
*/
int // ==> minimum motion cost after search
SubPelBlockMotionSearch (pel_t** orig_pic, // <-- original pixel values for the AxB block
int ref, // <-- reference frame (0... or -1 (backward))
int list, // <-- reference picture list
int pic_pix_x, // <-- absolute x-coordinate of regarded AxB block
int pic_pix_y, // <-- absolute y-coordinate of regarded AxB block
int blocktype, // <-- block type (1-16x16 ... 7-4x4)
int pred_mv_x, // <-- motion vector predictor (x) in sub-pel units
int pred_mv_y, // <-- motion vector predictor (y) in sub-pel units
int* mv_x, // <--> in: search center (x) / out: motion vector (x) - in pel units
int* mv_y, // <--> in: search center (y) / out: motion vector (y) - in pel units
int search_pos2, // <-- search positions for half-pel search (default: 9)
int search_pos4, // <-- search positions for quarter-pel search (default: 9)
int min_mcost, // <-- minimum motion cost (cost for center or huge value)
double lambda // <-- lagrangian parameter for determining motion cost
)
{
int diff[16], *d;
int pos, best_pos, mcost, abort_search;
int y0, x0, ry0, rx0, ry;
int cand_mv_x, cand_mv_y;
int max_pos_x4, max_pos_y4;
pel_t *orig_line;
pel_t **ref_pic;
StorablePicture *ref_picture;
int lambda_factor = LAMBDA_FACTOR (lambda);
int mv_shift = 0;
int check_position0 = (blocktype==1 && *mv_x==0 && *mv_y==0 && input->hadamard && !input->rdopt && img->type!=B_SLICE && ref==0);
int blocksize_x = input->blc_size[blocktype][0];
int blocksize_y = input->blc_size[blocktype][1];
int pic4_pix_x = (pic_pix_x << 2);
int pic4_pix_y = (pic_pix_y << 2);
int min_pos2 = (input->hadamard ? 0 : 1);
int max_pos2 = (input->hadamard ? max(1,search_pos2) : search_pos2);
int list_offset = ((img->MbaffFrameFlag)&&(img->mb_data[img->current_mb_nr].mb_field))? img->current_mb_nr%2 ? 4 : 2 : 0;
int apply_weights = ( (active_pps->weighted_pred_flag && (img->type == P_SLICE || img->type == SP_SLICE)) ||
(active_pps->weighted_bipred_idc && (img->type == B_SLICE)));
int img_width, img_height;
ref_picture = listX[list+list_offset][ref];
if (apply_weights)
{
ref_pic = listX[list+list_offset][ref]->imgY_ups_w;
}
else
ref_pic = listX[list+list_offset][ref]->imgY_ups;
img_width = ref_picture->size_x;
img_height = ref_picture->size_y;
max_pos_x4 = ((ref_picture->size_x - blocksize_x+1)<<2);
max_pos_y4 = ((ref_picture->size_y - blocksize_y+1)<<2);
/*********************************
***** *****
***** HALF-PEL REFINEMENT *****
***** *****
*********************************/
//===== convert search center to quarter-pel units =====
*mv_x <<= 2;
*mv_y <<= 2;
//===== set function for getting pixel values =====
if ((pic4_pix_x + *mv_x > 1) && (pic4_pix_x + *mv_x < max_pos_x4 - 2) &&
(pic4_pix_y + *mv_y > 1) && (pic4_pix_y + *mv_y < max_pos_y4 - 2) )
{
PelY_14 = FastPelY_14;
}
else
{
PelY_14 = UMVPelY_14;
}
//===== loop over search positions =====
for (best_pos = 0, pos = min_pos2; pos < max_pos2; pos++)
{
cand_mv_x = *mv_x + (spiral_search_x[pos] << 1); // quarter-pel units
cand_mv_y = *mv_y + (spiral_search_y[pos] << 1); // quarter-pel units
//----- set motion vector cost -----
mcost = MV_COST (lambda_factor, mv_shift, cand_mv_x, cand_mv_y, pred_mv_x, pred_mv_y);
if (check_position0 && pos==0)
{
mcost -= WEIGHTED_COST (lambda_factor, 16);
}
if (mcost >= min_mcost) continue;
//----- add up SATD -----
for (y0=0, abort_search=0; y0<blocksize_y && !abort_search; y0+=4)
{
ry0 = ((pic_pix_y+y0)<<2) + cand_mv_y;
for (x0=0; x0<blocksize_x; x0+=4)
{
rx0 = ((pic_pix_x+x0)<<2) + cand_mv_x;
d = diff;
orig_line = orig_pic [y0 ]; ry=ry0;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d++ = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
orig_line = orig_pic [y0+1]; ry=ry0+4;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d++ = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
orig_line = orig_pic [y0+2]; ry=ry0+8;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d++ = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
orig_line = orig_pic [y0+3]; ry=ry0+12;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
if ((mcost += SATD (diff, input->hadamard)) > min_mcost)
{
abort_search = 1;
break;
}
}
}
if (mcost < min_mcost)
{
min_mcost = mcost;
best_pos = pos;
}
}
if (best_pos)
{
*mv_x += (spiral_search_x [best_pos] << 1);
*mv_y += (spiral_search_y [best_pos] << 1);
}
/************************************
***** *****
***** QUARTER-PEL REFINEMENT *****
***** *****
************************************/
//===== set function for getting pixel values =====
if ((pic4_pix_x + *mv_x > 1) && (pic4_pix_x + *mv_x < max_pos_x4 - 1) &&
(pic4_pix_y + *mv_y > 1) && (pic4_pix_y + *mv_y < max_pos_y4 - 1) )
{
PelY_14 = FastPelY_14;
}
else
{
PelY_14 = UMVPelY_14;
}
//===== loop over search positions =====
for (best_pos = 0, pos = 1; pos < search_pos4; pos++)
{
cand_mv_x = *mv_x + spiral_search_x[pos]; // quarter-pel units
cand_mv_y = *mv_y + spiral_search_y[pos]; // quarter-pel units
//----- set motion vector cost -----
mcost = MV_COST (lambda_factor, mv_shift, cand_mv_x, cand_mv_y, pred_mv_x, pred_mv_y);
if (mcost >= min_mcost) continue;
//----- add up SATD -----
for (y0=0, abort_search=0; y0<blocksize_y && !abort_search; y0+=4)
{
ry0 = ((pic_pix_y+y0)<<2) + cand_mv_y;
for (x0=0; x0<blocksize_x; x0+=4)
{
rx0 = ((pic_pix_x+x0)<<2) + cand_mv_x;
d = diff;
orig_line = orig_pic [y0 ]; ry=ry0;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d++ = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
orig_line = orig_pic [y0+1]; ry=ry0+4;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d++ = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
orig_line = orig_pic [y0+2]; ry=ry0+8;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d++ = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
orig_line = orig_pic [y0+3]; ry=ry0+12;
*d++ = orig_line[x0 ] - PelY_14 (ref_pic, ry, rx0 , img_height, img_width);
*d++ = orig_line[x0+1] - PelY_14 (ref_pic, ry, rx0+ 4, img_height, img_width);
*d++ = orig_line[x0+2] - PelY_14 (ref_pic, ry, rx0+ 8, img_height, img_width);
*d = orig_line[x0+3] - PelY_14 (ref_pic, ry, rx0+12, img_height, img_width);
if ((mcost += SATD (diff, input->hadamard)) > min_mcost)
{
abort_search = 1;
break;
}
}
}
if (mcost < min_mcost)
{
min_mcost = mcost;
best_pos = pos;
}
}
if (best_pos)
{
*mv_x += spiral_search_x [best_pos];
*mv_y += spiral_search_y [best_pos];
}
//===== return minimum motion cost =====
return min_mcost;
}
/*!
***********************************************************************
* \brief
* Block motion search
***********************************************************************
*/
int //!< minimum motion cost after search
BlockMotionSearch (int ref, //!< reference idx
int list, //!< reference pciture list
int mb_x, //!< x-coordinate inside macroblock
int mb_y, //!< y-coordinate inside macroblock
int blocktype, //!< block type (1-16x16 ... 7-4x4)
int search_range, //!< 1-d search range for integer-position search
double lambda //!< lagrangian parameter for determining motion cost
)
{
static pel_t orig_val [256];
static pel_t *orig_pic [16] = {orig_val, orig_val+ 16, orig_val+ 32, orig_val+ 48,
orig_val+ 64, orig_val+ 80, orig_val+ 96, orig_val+112,
orig_val+128, orig_val+144, orig_val+160, orig_val+176,
orig_val+192, orig_val+208, orig_val+224, orig_val+240};
int pred_mv_x, pred_mv_y, mv_x, mv_y, i, j;
int max_value = (1<<20);
int min_mcost = max_value;
int block_x = (mb_x>>2);
int block_y = (mb_y>>2);
int bsx = input->blc_size[blocktype][0];
int bsy = input->blc_size[blocktype][1];
int pic_pix_x = img->opix_x + mb_x;
int pic_pix_y = img->opix_y + mb_y;
int* pred_mv;
int*** mv_array = enc_picture->mv[list];
int****** all_mv = img->all_mv;
#ifdef WIN32
struct _timeb tstruct1;
struct _timeb tstruct2;
#else
struct timeb tstruct1;
struct timeb tstruct2;
#endif
int me_tmp_time;
int N_Bframe=0, n_Bframe=0;
if(input->FMEnable)
{
N_Bframe = input->successive_Bframe;
n_Bframe =(N_Bframe) ? ((Bframe_ctr%N_Bframe)+1) : 0 ;
}
pred_mv = img->pred_mv[block_x][block_y][list][ref][blocktype];
//==================================
//===== GET ORIGINAL BLOCK =====
//==================================
for (j = 0; j < bsy; j++)
{
for (i = 0; i < bsx; i++)
{
orig_pic[j][i] = imgY_org[pic_pix_y+j][pic_pix_x+i];
}
}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -