快速快速双线性插值

图像缩放是最常用的图像操作，最近实现了一次双线性插值，适用于单通道8位灰度图像，可以同时处理缩小，放大。

加速手段有三个：

1：使用可分离方式，先后处理行和列

2：构建缓冲区，以避免对某些行重复插值。

3：使用整形运算，避免浮点运算。

performace:

测试图像： C:\Documents and Settings\All Users\Documents\My Pictures\示例图片\Water lilies.jpg

scale	_ieInterpImageBilinear8UC1（ms）	cvResize（ms）	ratio
0.3	346	390	89%
0.5	939	1047	90%
0.8	1671	1507	110%
1.2	2885	1943	148%
1.5	3951	2678	150%
2	6119	3642	170%

上表都是执行1000次的总时间，可以看出，在缩小时速度比opencv的cvResize相当，在放大时，不幸的比cvResize慢上50%。贴出代码求解。

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90

typedef struct _Image{      int w;      int h;      int c;      int step;      int type;      void * data; } Image;   static void _ieInterpImageBilinear8UC1_RowFilter(unsigned char * src, long * dst, int len, int * leftIdx, int * rightIdx, long * weight, int shift) {      int i, j;      for (i = 0, j = 1; i < len - 2; i+=2, j+=2){          dst[i] = ((1<<shift) - weight[i])*src[leftIdx[i]] + weight[i]*src[rightIdx[i]];          dst[j] = ((1<<shift) - weight[j])*src[leftIdx[j]] + weight[j]*src[rightIdx[j]];        }      for ( ; j < len; j++){          dst[j] = ((1<<shift) - weight[j])*src[leftIdx[j]] + weight[j]*src[rightIdx[j]];      } }   static void _ieInterpImageBilinear8UC1(Image* src, Image* dst) {      int i, j;      int sw, sh, sstep;      int dw, dh, dstep;      unsigned char *sdata, *ddata;      float horScaleRatio, verScaleRatio;      long *rowBuf1, *rowBuf2;      long *upLinePtr, *downLinePtr, *tempPtr;      long *horWeight;      int *horLeftIdx, *horRightIdx;      int preVerUpIdx, preVerDownIdx;      int shift = 8;        sw = src->w; sh = src->h; sstep = src->step; sdata = (unsigned char *)(src->data);      dw = dst->w; dh = dst->h; dstep = dst->step; ddata = (unsigned char *)(dst->data);        horScaleRatio = sw / ( float )(dw);      verScaleRatio = sh / ( float )(dh);        rowBuf1 = new long [dw];      rowBuf2 = new long [dw];      horWeight = new long [dw];      horLeftIdx = new int [dw];      horRightIdx = new int [dw];        //col interpolation      //计算目标图像像素横向的左右邻居序号，和权重。      for (i = 0; i < dw; i++){          float pos = (i + 0.5f) * horScaleRatio;          horLeftIdx[i] = ( int )(IET_MAX(pos - 0.5f, 0));          horRightIdx[i] = ( int )(IET_MIN(pos + 0.5f, sw-1));          horWeight[i] = ( long ) ( fabs (pos - 0.5f - horLeftIdx[i]) * (1<<shift));      }        preVerUpIdx = -1;      preVerDownIdx = -1;      upLinePtr = rowBuf1;      downLinePtr = rowBuf2;      for (j = 0; j < dh; j++){                 float pos = (j + 0.5f) * verScaleRatio;          int verUpIdx = ( int )(IET_MAX(pos - 0.5f, 0));          int verDownIdx = ( int )(IET_MIN(pos + 0.5f, sh-1));          long verWeight = ( long ) ( fabs (pos - 0.5f - verUpIdx) * (1<<shift));            if (verUpIdx == preVerUpIdx && verDownIdx == preVerDownIdx){              ; //do nothing          }          else if (verUpIdx == preVerDownIdx){              IET_SWAP(upLinePtr, downLinePtr, tempPtr);              _ieInterpImageBilinear8UC1_RowFilter(sdata + sstep*verDownIdx, downLinePtr, dw, horLeftIdx, horRightIdx, horWeight, shift);          } else {              _ieInterpImageBilinear8UC1_RowFilter(sdata + sstep*verUpIdx,   upLinePtr, dw, horLeftIdx, horRightIdx, horWeight, shift);              _ieInterpImageBilinear8UC1_RowFilter(sdata + sstep*verDownIdx, downLinePtr, dw, horLeftIdx, horRightIdx, horWeight, shift);          }                   for (i = 0; i < dw; i++){              ddata[dstep*j + i] = (unsigned char ) ( (((1<<shift) - verWeight)*upLinePtr[i] + verWeight*downLinePtr[i]) >> (2*shift) );          }          preVerUpIdx = verUpIdx;          preVerDownIdx = verDownIdx;      }      delete []rowBuf1;      delete []rowBuf2;      delete []horWeight;      delete []horLeftIdx;      delete []horRightIdx; }

使用神奇的四次展开循环，速度提升15%+。
code如下。

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90

static void _ieInterpImageBilinear8UC1_Ver3_RowFilter(unsigned char * src, long * dst, int len, int * leftIdx, int * rightIdx, long * weight, int shift) {      int i;      for (i = 0; i < len - 4; i+=4){          *dst++ = ((1<<shift) - weight[i])*src[leftIdx[i]] + weight[i]*src[rightIdx[i]];          *dst++ = ((1<<shift) - weight[i+1])*src[leftIdx[i+1]] + weight[i+1]*src[rightIdx[i+1]];          *dst++ = ((1<<shift) - weight[i+2])*src[leftIdx[i+2]] + weight[i+2]*src[rightIdx[i+2]];          *dst++ = ((1<<shift) - weight[i+3])*src[leftIdx[i+3]] + weight[i+3]*src[rightIdx[i+3]];        }      for ( ; i < len; ++i){          *dst++ = ((1<<shift) - weight[i])*src[leftIdx[i]] + weight[i]*src[rightIdx[i]];      } }   static void _ieInterpImageBilinear8UC1_Ver3(Image* src, Image* dst) {      int i, j;      int sw, sh, sstep;      int dw, dh, dstep;      unsigned char *sdata, *ddata;      float horScaleRatio, verScaleRatio;      long *rowBuf1, *rowBuf2;      long *upLinePtr, *downLinePtr, *tempPtr;      long *horWeight;      int *horLeftIdx, *horRightIdx;      int preVerUpIdx, preVerDownIdx;      int shift = 8;        sw = src->w; sh = src->h; sstep = src->step; sdata = (unsigned char *)(src->data);      dw = dst->w; dh = dst->h; dstep = dst->step; ddata = (unsigned char *)(dst->data);        horScaleRatio = sw / ( float )(dw);      verScaleRatio = sh / ( float )(dh);        rowBuf1 = new long [dw];      rowBuf2 = new long [dw];      horWeight = new long [dw];      horLeftIdx = new int [dw];      horRightIdx = new int [dw];        //col interpolation      //计算目标图像像素横向的左右邻居序号，和权重。      for (i = 0; i < dw; i++){          float pos = (i + 0.5f) * horScaleRatio;          horLeftIdx[i] = ( int )(IET_MAX(pos - 0.5f, 0));          horRightIdx[i] = ( int )(IET_MIN(pos + 0.5f, sw-1));          horWeight[i] = ( long ) ( fabs (pos - 0.5f - horLeftIdx[i]) * (1<<shift));      }        preVerUpIdx = -1;      preVerDownIdx = -1;      upLinePtr = rowBuf1;      downLinePtr = rowBuf2;      for (j = 0; j < dh; j++){                 float pos = (j + 0.5f) * verScaleRatio;          int verUpIdx = ( int )(IET_MAX(pos - 0.5f, 0));          int verDownIdx = ( int )(IET_MIN(pos + 0.5f, sh-1));          long verWeight = ( long ) ( fabs (pos - 0.5f - verUpIdx) * (1<<shift));            if (verUpIdx == preVerUpIdx && verDownIdx == preVerDownIdx){              ; //do nothing          }          else if (verUpIdx == preVerDownIdx){              IET_SWAP(upLinePtr, downLinePtr, tempPtr);              _ieInterpImageBilinear8UC1_Ver3_RowFilter(sdata + sstep*verDownIdx, downLinePtr, dw, horLeftIdx, horRightIdx, horWeight, shift);          } else {              _ieInterpImageBilinear8UC1_Ver3_RowFilter(sdata + sstep*verUpIdx,   upLinePtr, dw, horLeftIdx, horRightIdx, horWeight, shift);              _ieInterpImageBilinear8UC1_Ver3_RowFilter(sdata + sstep*verDownIdx, downLinePtr, dw, horLeftIdx, horRightIdx, horWeight, shift);          }                   unsigned char * _ptr = ddata + dstep*j;          for (i = 0; i < dw-4; i+=4){              *_ptr++ = (unsigned char ) ( (((1<<shift) - verWeight)*upLinePtr[i] + verWeight*downLinePtr[i]) >> (2*shift) );              *_ptr++ = (unsigned char ) ( (((1<<shift) - verWeight)*upLinePtr[i+1] + verWeight*downLinePtr[i+1]) >> (2*shift) );              *_ptr++ = (unsigned char ) ( (((1<<shift) - verWeight)*upLinePtr[i+2] + verWeight*downLinePtr[i+2]) >> (2*shift) );              *_ptr++ = (unsigned char ) ( (((1<<shift) - verWeight)*upLinePtr[i+3] + verWeight*downLinePtr[i+3]) >> (2*shift) );          }          for (; i < dw; i++){              *_ptr++ = (unsigned char ) ( (((1<<shift) - verWeight)*upLinePtr[i] + verWeight*downLinePtr[i]) >> (2*shift) );          }          preVerUpIdx