/*
* Copyright (c) 1998, 2011, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
/*
* FUNCTION
* mlib_ImageAffineEdgeZero - implementation of MLIB_EDGE_DST_FILL_ZERO
* edge condition
* mlib_ImageAffineEdgeNearest - implementation of MLIB_EDGE_OP_NEAREST
* edge condition
* void mlib_ImageAffineEdgeExtend_BL - implementation of MLIB_EDGE_SRC_EXTEND
* edge condition for MLIB_BILINEAR filter
* void mlib_ImageAffineEdgeExtend_BC - implementation of MLIB_EDGE_SRC_EXTEND
* edge condition for MLIB_BICUBIC filter
* void mlib_ImageAffineEdgeExtend_BC2 - implementation of MLIB_EDGE_SRC_EXTEND
* edge condition for MLIB_BICUBIC2 filter
*
* DESCRIPTION
* mlib_ImageAffineEdgeZero:
* This function fills the edge pixels (i.e. thouse one which can not
* be interpolated with given resampling filter because their prototypes
* in the source image lie too close to the border) in the destination
* image with zeroes.
*
* mlib_ImageAffineEdgeNearest:
* This function fills the edge pixels (i.e. thouse one which can not
* be interpolated with given resampling filter because their prototypes
* in the source image lie too close to the border) in the destination
* image according to the nearest neighbour interpolation.
*
* mlib_ImageAffineEdgeExtend_BL:
* This function fills the edge pixels (i.e. thouse one which can not
* be interpolated with given resampling filter because their prototypes
* in the source image lie too close to the border) in the destination
* image according to the bilinear interpolation with border pixels extend
* of source image.
*
* mlib_ImageAffineEdgeExtend_BC:
* This function fills the edge pixels (i.e. thouse one which can not
* be interpolated with given resampling filter because their prototypes
* in the source image lie too close to the border) in the destination
* image according to the bicubic interpolation with border pixels extend
* of source image.
*
* mlib_ImageAffineEdgeExtend_BC2:
* This function fills the edge pixels (i.e. thouse one which can not
* be interpolated with given resampling filter because their prototypes
* in the source image lie too close to the border) in the destination
* image according to the bicubic2 interpolation with border pixels extend
* of source image.
*/
#include "mlib_image.h"
#include "mlib_ImageAffine.h"
/***************************************************************/
#define FLT_SHIFT_U8 4
#define FLT_MASK_U8 (((1 << 8) - 1) << 4)
#define FLT_SHIFT_S16 3
#define FLT_MASK_S16 (((1 << 9) - 1) << 4)
#define MLIB_SIGN_SHIFT 31
/***************************************************************/
#define D64mlib_u8(X) mlib_U82D64[X]
#define D64mlib_s16(X) ((mlib_d64)(X))
#define D64mlib_u16(X) ((mlib_d64)(X))
#define D64mlib_s32(X) ((mlib_d64)(X))
#define D64mlib_f32(X) ((mlib_d64)(X))
#define D64mlib_d64(X) ((mlib_d64)(X))
/***************************************************************/
#ifdef MLIB_USE_FTOI_CLAMPING
#define SATmlib_u8(DST, val0) \
DST = ((mlib_s32)(val0 - sat) >> 24) ^ 0x80
#define SATmlib_s16(DST, val0) \
DST = ((mlib_s32)val0) >> 16
#define SATmlib_u16(DST, val0) \
DST = ((mlib_s32)(val0 - sat) >> 16) ^ 0x8000
#define SATmlib_s32(DST, val0) \
DST = val0
#else
#define SATmlib_u8(DST, val0) \
val0 -= sat; \
if (val0 >= MLIB_S32_MAX) \
val0 = MLIB_S32_MAX; \
if (val0 <= MLIB_S32_MIN) \
val0 = MLIB_S32_MIN; \
DST = ((mlib_s32) val0 >> 24) ^ 0x80
#define SATmlib_s16(DST, val0) \
if (val0 >= MLIB_S32_MAX) \
val0 = MLIB_S32_MAX; \
if (val0 <= MLIB_S32_MIN) \
val0 = MLIB_S32_MIN; \
DST = (mlib_s32)val0 >> 16
#define SATmlib_u16(DST, val0) \
val0 -= sat; \
if (val0 >= MLIB_S32_MAX) \
val0 = MLIB_S32_MAX; \
if (val0 <= MLIB_S32_MIN) \
val0 = MLIB_S32_MIN; \
DST = ((mlib_s32)val0 >> 16) ^ 0x8000
#define SATmlib_s32(DST, val0) \
if (val0 >= MLIB_S32_MAX) \
val0 = MLIB_S32_MAX; \
if (val0 <= MLIB_S32_MIN) \
val0 = MLIB_S32_MIN; \
DST = (mlib_s32)val0
#endif
/***************************************************************/
#define SATmlib_f32(DST, val0) \
DST = (mlib_f32)val0
/***************************************************************/
#define SATmlib_d64(DST, val0) \
DST = val0
/***************************************************************/
#define MLIB_EDGE_ZERO_LINE(TYPE, Left, Right) \
dp = (TYPE*)data + channels * Left; \
dstLineEnd = (TYPE*)data + channels * Right; \
\
for (; dp < dstLineEnd; dp++) { \
*dp = zero; \
}
/***************************************************************/
#define MLIB_EDGE_NEAREST_LINE(TYPE, Left, Right) \
dp = (TYPE*)data + channels * Left; \
size = Right - Left; \
\
for (j = 0; j < size; j++) { \
ySrc = Y >> MLIB_SHIFT; \
xSrc = X >> MLIB_SHIFT; \
sp = (TYPE*)lineAddr[ySrc] + xSrc * channels; \
\
for (k = 0; k < channels; k++) dp[k] = sp[k]; \
\
Y += dY; \
X += dX; \
dp += channels; \
}
/***************************************************************/
#define MLIB_EDGE_BL(TYPE, Left, Right) \
dp = (TYPE*)data + channels * Left; \
size = Right - Left; \
\
for (j = 0; j < size; j++) { \
ySrc = ((Y - 32768) >> MLIB_SHIFT); \
xSrc = ((X - 32768) >> MLIB_SHIFT); \
\
t = ((X - 32768) & MLIB_MASK) * scale; \
u = ((Y - 32768) & MLIB_MASK) * scale; \
\
xDelta = (((xSrc + 1 - srcWidth )) >> MLIB_SIGN_SHIFT) & channels; \
yDelta = (((ySrc + 1 - srcHeight)) >> MLIB_SIGN_SHIFT) & srcStride; \
\
xFlag = (xSrc >> (MLIB_SIGN_SHIFT - MLIB_SHIFT)); \
xSrc = xSrc + (1 & xFlag); \
xDelta = xDelta &~ xFlag; \
\
yFlag = (ySrc >> (MLIB_SIGN_SHIFT - MLIB_SHIFT)); \
ySrc = ySrc + (1 & yFlag); \
yDelta = yDelta &~ yFlag; \
\
sp = (TYPE*)lineAddr[ySrc] + xSrc * channels; \
\
for (k = 0; k < channels; k++) { \
a00 = D64##TYPE(sp[0]); \
a01 = D64##TYPE(sp[xDelta]); \
a10 = D64##TYPE(sp[yDelta]); \
a11 = D64##TYPE(sp[yDelta + xDelta]); \
pix0 = (a00 * (1 - t) + a01 * t) * (1 - u) + \
(a10 * (1 - t) + a11 * t) * u; \
\
dp[k] = (TYPE)pix0; \
sp++; \
} \
\
X += dX; \
Y += dY; \
dp += channels; \
}
/***************************************************************/
#define GET_FLT_TBL(X, xf0, xf1, xf2, xf3) \
filterpos = ((X - 32768) >> flt_shift) & flt_mask; \
fptr = (mlib_f32 *) ((mlib_u8 *)flt_tbl + filterpos); \
\
xf0 = fptr[0]; \
xf1 = fptr[1]; \
xf2 = fptr[2]; \
xf3 = fptr[3]
/***************************************************************/
#define GET_FLT_BC(X, xf0, xf1, xf2, xf3) \
dx = ((X - 32768) & MLIB_MASK) * scale; \
dx_2 = 0.5 * dx; \
dx2 = dx * dx; \
dx3_2 = dx_2 * dx2; \
dx3_3 = 3.0 * dx3_2; \
\
xf0 = dx2 - dx3_2 - dx_2; \
xf1 = dx3_3 - 2.5 * dx2 + 1.0; \
xf2 = 2.0 * dx2 - dx3_3 + dx_2; \
xf3 = dx3_2 - 0.5 * dx2
/***************************************************************/
#define GET_FLT_BC2(X, xf0, xf1, xf2, xf3) \
dx = ((X - 32768) & MLIB_MASK) * scale; \
dx2 = dx * dx; \
dx3_2 = dx * dx2; \
dx3_3 = 2.0 * dx2; \
\
xf0 = - dx3_2 + dx3_3 - dx; \
xf1 = dx3_2 - dx3_3 + 1.0; \
xf2 = - dx3_2 + dx2 + dx; \
xf3 = dx3_2 - dx2
/***************************************************************/
#define CALC_SRC_POS(X, Y, channels, srcStride) \
xSrc = ((X - 32768) >> MLIB_SHIFT); \
ySrc = ((Y - 32768) >> MLIB_SHIFT); \
\
xDelta0 = ((~((xSrc - 1) >> MLIB_SIGN_SHIFT)) & (- channels)); \
yDelta0 = ((~((ySrc - 1) >> MLIB_SIGN_SHIFT)) & (- srcStride)); \
xDelta1 = ((xSrc + 1 - srcWidth) >> MLIB_SIGN_SHIFT) & (channels); \
yDelta1 = ((ySrc + 1 - srcHeight) >> MLIB_SIGN_SHIFT) & (srcStride); \
xDelta2 = xDelta1 + (((xSrc + 2 - srcWidth) >> MLIB_SIGN_SHIFT) & (channels)); \
yDelta2 = yDelta1 + (((ySrc + 2 - srcHeight) >> MLIB_SIGN_SHIFT) & (srcStride)); \
\
xFlag = (xSrc >> (MLIB_SIGN_SHIFT - MLIB_SHIFT)); \
xSrc = xSrc + (1 & xFlag); \
xDelta2 -= (xDelta1 & xFlag); \
xDelta1 = (xDelta1 &~ xFlag); \
\
yFlag = (ySrc >> (MLIB_SIGN_SHIFT - MLIB_SHIFT)); \
ySrc = ySrc + (1 & yFlag); \
yDelta2 -= (yDelta1 & yFlag); \
yDelta1 = yDelta1 &~ yFlag
/***************************************************************/
#define MLIB_EDGE_BC_LINE(TYPE, Left, Right, GET_FILTER) \
dp = (TYPE*)data + channels * Left; \
size = Right - Left; \
\
for (j = 0; j < size; j++) { \
GET_FILTER(X, xf0, xf1, xf2, xf3); \
GET_FILTER(Y, yf0, yf1, yf2, yf3); \
\
CALC_SRC_POS(X, Y, channels, srcStride); \
\
sp = (TYPE*)lineAddr[ySrc] + channels*xSrc; \
\
for (k = 0; k < channels; k++) { \
c0 = D64##TYPE(sp[yDelta0 + xDelta0]) * xf0 + \
D64##TYPE(sp[yDelta0 ]) * xf1 + \
D64##TYPE(sp[yDelta0 + xDelta1]) * xf2 + \
D64##TYPE(sp[yDelta0 + xDelta2]) * xf3; \
\
c1 = D64##TYPE(sp[xDelta0]) * xf0 + \
D64##TYPE(sp[ 0]) * xf1 + \
D64##TYPE(sp[xDelta1]) * xf2 + \
D64##TYPE(sp[xDelta2]) * xf3; \
\
c2 = D64##TYPE(sp[yDelta1 + xDelta0]) * xf0 + \
D64##TYPE(sp[yDelta1 ]) * xf1 + \
D64##TYPE(sp[yDelta1 + xDelta1]) * xf2 + \
D64##TYPE(sp[yDelta1 + xDelta2]) * xf3; \
\
c3 = D64##TYPE(sp[yDelta2 + xDelta0]) * xf0 + \
D64##TYPE(sp[yDelta2 ]) * xf1 + \
D64##TYPE(sp[yDelta2 + xDelta1]) * xf2 + \
D64##TYPE(sp[yDelta2 + xDelta2]) * xf3; \
\
val0 = c0*yf0 + c1*yf1 + c2*yf2 + c3*yf3; \
\
SAT##TYPE(dp[k], val0); \
\
sp++; \
} \
\
X += dX; \
Y += dY; \
dp += channels; \
}
/***************************************************************/
#define MLIB_EDGE_BC_TBL(TYPE, Left, Right) \
MLIB_EDGE_BC_LINE(TYPE, Left, Right, GET_FLT_TBL)
/***************************************************************/
#define MLIB_EDGE_BC(TYPE, Left, Right) \
MLIB_EDGE_BC_LINE(TYPE, Left, Right, GET_FLT_BC)
/***************************************************************/
#define MLIB_EDGE_BC2(TYPE, Left, Right) \
MLIB_EDGE_BC_LINE(TYPE, Left, Right, GET_FLT_BC2)
/***************************************************************/
#define MLIB_PROCESS_EDGES_ZERO(TYPE) { \
TYPE *dp, *dstLineEnd; \
\
for (i = yStartE; i < yStart; i++) { \
xLeftE = leftEdgesE[i]; \
xRightE = rightEdgesE[i] + 1; \
data += dstStride; \
\
MLIB_EDGE_ZERO_LINE(TYPE, xLeftE, xRightE); \
} \
\
for (; i <= yFinish; i++) { \
xLeftE = leftEdgesE[i]; \
xRightE = rightEdgesE[i] + 1; \
xLeft = leftEdges[i]; \
xRight = rightEdges[i] + 1; \
data += dstStride; \
\
if (xLeft < xRight) { \
MLIB_EDGE_ZERO_LINE(TYPE, xLeftE, xLeft); \
} else { \
xRight = xLeftE; \
} \
\
MLIB_EDGE_ZERO_LINE(TYPE, xRight, xRightE); \
} \
\
for (; i <= yFinishE; i++) { \
xLeftE = leftEdgesE[i]; \
xRightE = rightEdgesE[i] + 1; \
data += dstStride; \
\
MLIB_EDGE_ZERO_LINE(TYPE, xLeftE, xRightE); \
} \
}
/***************************************************************/
#define MLIB_PROCESS_EDGES(PROCESS_LINE, TYPE) { \
TYPE *sp, *dp; \
mlib_s32 k, size; \
\
for (i = yStartE; i < yStart; i++) { \
xLeftE = leftEdgesE[i]; \
xRightE = rightEdgesE[i] + 1; \
X = xStartsE[i]; \
Y = yStartsE[i]; \
data += dstStride; \
\
PROCESS_LINE(TYPE, xLeftE, xRightE); \
} \
\
for (; i <= yFinish; i++) { \
xLeftE = leftEdgesE[i]; \
xRightE = rightEdgesE[i] + 1; \
xLeft = leftEdges[i]; \
xRight = rightEdges[i] + 1; \
X = xStartsE[i]; \
Y = yStartsE[i]; \
data += dstStride; \
\
if (xLeft < xRight) { \
PROCESS_LINE(TYPE, xLeftE, xLeft); \
} else { \
xRight = xLeftE; \
} \
\
X = xStartsE[i] + dX * (xRight - xLeftE); \
Y = yStartsE[i] + dY * (xRight - xLeftE); \
PROCESS_LINE(TYPE, xRight, xRightE); \
} \
\
for (; i <= yFinishE; i++) { \
xLeftE = leftEdgesE[i]; \
xRightE = rightEdgesE[i] + 1; \
X = xStartsE[i]; \
Y = yStartsE[i]; \
data += dstStride; \
\
PROCESS_LINE(TYPE, xLeftE, xRightE); \
} \
}
/***************************************************************/
#define GET_EDGE_PARAMS_ZERO() \
mlib_image *dst = param -> dst; \
mlib_s32 *leftEdges = param -> leftEdges; \
mlib_s32 *rightEdges = param -> rightEdges; \
mlib_s32 *leftEdgesE = param_e -> leftEdges; \
mlib_s32 *rightEdgesE = param_e -> rightEdges; \
mlib_type type = mlib_ImageGetType(dst); \
mlib_s32 channels = mlib_ImageGetChannels(dst); \
mlib_s32 dstStride = mlib_ImageGetStride(dst); \
mlib_s32 yStart = param -> yStart; \
mlib_s32 yFinish = param -> yFinish; \
mlib_s32 yStartE = param_e -> yStart; \
mlib_s32 yFinishE = param_e -> yFinish; \
mlib_u8 *data = param_e -> dstData; \
mlib_s32 xLeft, xRight, xLeftE, xRightE; \
mlib_s32 i
/***************************************************************/
#define GET_EDGE_PARAMS_NN() \
GET_EDGE_PARAMS_ZERO(); \
mlib_s32 *xStartsE = param_e -> xStarts; \
mlib_s32 *yStartsE = param_e -> yStarts; \
mlib_u8 **lineAddr = param -> lineAddr; \
mlib_s32 dX = param_e -> dX; \
mlib_s32 dY = param_e -> dY; \
mlib_s32 xSrc, ySrc, X, Y; \
mlib_s32 j
/***************************************************************/
#define GET_EDGE_PARAMS() \
GET_EDGE_PARAMS_NN(); \
mlib_image *src = param -> src; \
mlib_s32 srcWidth = mlib_ImageGetWidth(src); \
mlib_s32 srcHeight = mlib_ImageGetHeight(src); \
mlib_s32 srcStride = mlib_ImageGetStride(src)
/***************************************************************/
void mlib_ImageAffineEdgeZero(mlib_affine_param *param,
mlib_affine_param *param_e)
{
GET_EDGE_PARAMS_ZERO();
mlib_s32 zero = 0;
switch (type) {
case MLIB_BYTE:
MLIB_PROCESS_EDGES_ZERO(mlib_u8);
break;
case MLIB_SHORT:
case MLIB_USHORT:
MLIB_PROCESS_EDGES_ZERO(mlib_s16);
break;
case MLIB_INT:
case MLIB_FLOAT:
MLIB_PROCESS_EDGES_ZERO(mlib_s32);
break;
case MLIB_DOUBLE:{
mlib_d64 zero = 0;
MLIB_PROCESS_EDGES_ZERO(mlib_d64);
break;
}
default:
/* Image type MLIB_BIT is not used in java, so we can ignore it. */
break;
}
}
/***************************************************************/
void mlib_ImageAffineEdgeNearest(mlib_affine_param *param,
mlib_affine_param *param_e)
{
GET_EDGE_PARAMS_NN();
switch (type) {
case MLIB_BYTE:
MLIB_PROCESS_EDGES(MLIB_EDGE_NEAREST_LINE, mlib_u8);
break;
case MLIB_SHORT:
case MLIB_USHORT:
MLIB_PROCESS_EDGES(MLIB_EDGE_NEAREST_LINE, mlib_s16);
break;
case MLIB_INT:
case MLIB_FLOAT:
MLIB_PROCESS_EDGES(MLIB_EDGE_NEAREST_LINE, mlib_s32);
break;
case MLIB_DOUBLE:
MLIB_PROCESS_EDGES(MLIB_EDGE_NEAREST_LINE, mlib_d64);
break;
default:
/* Image type MLIB_BIT is not used in java, so we can ignore it. */
break;
}
}
/***************************************************************/
mlib_status mlib_ImageAffineEdgeExtend_BL(mlib_affine_param *param,
mlib_affine_param *param_e)
{
GET_EDGE_PARAMS();
mlib_d64 scale = 1.0 / (mlib_d64) MLIB_PREC;
mlib_s32 xDelta, yDelta, xFlag, yFlag;
mlib_d64 t, u, pix0;
mlib_d64 a00, a01, a10, a11;
switch (type) {
case MLIB_BYTE:
MLIB_PROCESS_EDGES(MLIB_EDGE_BL, mlib_u8);
/**代码未完, 请加载全部代码(NowJava.com).**/