2 * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved.
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice shall be included in
12 * all copies or substantial portions of the Software.
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
17 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
19 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
20 * DEALINGS IN THE SOFTWARE.
23 #include "cuda/vector_helpers.cuh"
26 __device__ inline void Subsample_Bilinear(cudaTextureObject_t tex,
28 int dst_width, int dst_height, int dst_pitch,
29 int src_width, int src_height,
32 int xo = blockIdx.x * blockDim.x + threadIdx.x;
33 int yo = blockIdx.y * blockDim.y + threadIdx.y;
35 if (yo < dst_height && xo < dst_width)
37 float hscale = (float)src_width / (float)dst_width;
38 float vscale = (float)src_height / (float)dst_height;
39 float xi = (xo + 0.5f) * hscale;
40 float yi = (yo + 0.5f) * vscale;
41 // 3-tap filter weights are {wh,1.0,wh} and {wv,1.0,wv}
42 float wh = min(max(0.5f * (hscale - 1.0f), 0.0f), 1.0f);
43 float wv = min(max(0.5f * (vscale - 1.0f), 0.0f), 1.0f);
44 // Convert weights to two bilinear weights -> {wh,1.0,wh} -> {wh,0.5,0} + {0,0.5,wh}
45 float dx = wh / (0.5f + wh);
46 float dy = wv / (0.5f + wv);
50 r += tex2D<T>(tex, xi - dx, yi - dy);
51 r += tex2D<T>(tex, xi + dx, yi - dy);
52 r += tex2D<T>(tex, xi - dx, yi + dy);
53 r += tex2D<T>(tex, xi + dx, yi + dy);
54 vec_set(dst[yo*dst_pitch+xo], r >> 2);
60 #define BILINEAR_KERNEL(T) \
61 __global__ void Subsample_Bilinear_ ## T(cudaTextureObject_t src_tex, \
63 int dst_width, int dst_height, int dst_pitch, \
64 int src_width, int src_height, \
67 Subsample_Bilinear<T>(src_tex, dst, \
68 dst_width, dst_height, dst_pitch, \
69 src_width, src_height, \
73 BILINEAR_KERNEL(uchar)
74 BILINEAR_KERNEL(uchar2)
75 BILINEAR_KERNEL(uchar4)
77 BILINEAR_KERNEL(ushort)
78 BILINEAR_KERNEL(ushort2)
79 BILINEAR_KERNEL(ushort4)