如何理解CS50 Problem Set4中edges滤镜的实现代码及运行结果？

这段代码用于完成CS50第四套作业的边缘滤镜拓展任务，核心逻辑是：先将原图像的RGBTRIPLE数据完整复制到临时数组，再遍历每个像素及其周围3×3邻域的像素，通过Sobel梯度算子（gx、gy）分别计算红、绿、蓝三个通道的梯度值，最后结合梯度值计算边缘强度并更新图像像素。

void edges(int height, int width, RGBTRIPLE image[height][width])
{
    RGBTRIPLE copy[height][width];

    long gxRed, gxGreen, gxBlue, gyRed, gyGreen, gyBlue;
    int row_x, row_y, column_x, column_y, a, b;
    int gx_arr[3][3] = {{-1, 0, 1}, {-2, 0, 2}, {-1, 0, 1}};
    int gy_arr[3][3] = {{-1, -2, -1}, {0, 0, 0}, {1, 2, 1}};

    for (int i = 0; i < height; i++)
    {
        for (int j = 0; j < width; j++)
        {
            copy[i][j] = image[i][j];
        }
    }

    for (int i = 0; i < height; i++)
    {
        for (int j = 0; j < width; j++)
        {
            gxRed = 0;
            gxGreen = 0;
            gxBlue = 0;
            gyRed = 0;
            gyGreen = 0;
            gyBlue = 0;

            for (a = i - 1, row_x = 0, row_y = 0; a <= i + 1; a++, row_x++, row_y++)
            {
                for (b = j - 1, column_x = 0, column_y = 0; b <= j + 1; b++, column_x++, column_y++)
                {
                    if (a < 0 || b < 0 || a > height || b > width)
                    {
                        continue;
                    }

                    gxRed += image[a][b].rgbtRed * gx_arr[row_x][column_x];
                    gxGreen += image[a][b].rgbtGreen * gx_arr[row_x][column_x];
                    gxBlue += image[a][b].rgbtBlue * gx_arr[row_x][column_x];

                    gyRed += image[a][b].rgbtRed * gy_arr[row_y][column_y];
                    gyGreen += image[a][b].rgbtGreen * gy_arr[row_y][column_y];
                    gyBlue += image[a][b].rgbtBlue * gy_arr[row_y][column_y];
                }
                column_x = 0;
                column_y = 0;
            }

            copy[i][j].rgbtRed   = fmin(round(sqrt(gxRed * gxRed + gyRed * gyRed)), 255);
            copy[i][j].rgbtGreen = fmin(round(sqrt(gxGreen * gxGreen + gyGreen * gyGreen)), 255);
            copy[i][j].rgbtBlue  = fmin(round(sqrt(gxBlue * gxBlue + gyBlue * gyBlue)), 255);
        }
    }

    for (int i = 0; i < height; i++)
    {
        for (int j = 0; j < width; j++)
        {
            image[i][j] = copy[i][j];
        }
    }

    return;
}

运行效果 #

• 测试效果1：原图经过处理后，物体的轮廓边缘被清晰提取，不同色彩区域的边界呈现高亮线条，符合边缘检测的预期。
• 测试效果2：另一组图像的处理结果显示，细节边缘保留完整，过渡自然，滤镜效果达标。

内容的提问来源于stack exchange，提问作者Rohan jain