写一个cuda做单通道图像双线性插值的代码

这是一段使用 CUDA 进行单通道图像双线性插值的代码示例:

#include <cuda_runtime.h>

__global__ void bilinear_interpolation(unsigned char *out, unsigned char *in, int w, int h, int new_w, int new_h) {
    int x = threadIdx.x + blockIdx.x * blockDim.x;
    int y = threadIdx.y + blockIdx.y * blockDim.y;
    if (x >= new_w || y >= new_h) return;
    
    float x_ratio = ((float)(w - 1)) / new_w;
    float y_ratio = ((float)(h - 1)) / new_h;
    float px = x * x_ratio;
    float py = y * y_ratio;
    int px1 = floor(px);
    int py1 = floor(py);
    int px2 = px1 < w - 1 ? px1 + 1 : px1;
    int py2 = py1 < h - 1 ? py1 + 1 : py1;
    float fx = px - px1;
    float fy = py - py1;
    float w1 = (1 - fx) * (1 - fy);
    float w2 = fx * (1 - fy);
    float w3 = (1 - fx) * fy;
    float w4 = fx * fy;
    int offset1 = py1 * w + px1;
    int offset2 = py1 * w + px2;
    int offset3 = py2 * w + px1;
    int offset4 = py2 * w + px2;

    int result = (w1 * in[offset1] + w2 * in[offset2] + w3 * in[offset3] + w4 * in[offset4]);
    out[y * new_w + x] = result;
}

int main() {
    // 初始化输入图像和输出图像
    int w = 100, h = 100;
    int new_w = 200, new_h = 200;
    unsigned char *in, *out;
    cudaMalloc(&in, w * h);
    cudaMalloc(&out, new_w * new_h);
    // 将输入图像数据复制到 GPU 上
    cudaMemcpy(in, in_data, w * h, cudaMemcpyHostToDevice);

    // 执行双线性插值
    dim3 block(16, 16);
    dim3 grid((new_w + block.x - 1) / block.x, (new_h + block.y - 1) / block.y);
    bilinear_interpolation<<<grid, block>>>(out, in, w, h, new_w