假设一幅大小为500 * 500的图像扩大1.5倍到750 * 750,创建一个750 * 750 的网格,使其与原图像间隔相同,然后缩小至原图大小,在原图中寻找最接近的像素(或周围的像素)进行赋值,最后再将结果放大
最邻近内插法寻找最近的像素赋值
双线性内插法v(x,y) = ax + by + cxy + d
双线性内插法参数计算
已知Q11, Q12, Q21, Q22,要插值的点为P点,首先在x轴上,对R1,R2两个点进行插值
然后根据R1和R2对P点进行插值
化简得
对于边界值的处理,若x1 < 0 ,则直接令f(Q11), f(Q12) = 0
处理结果 原图 扩大为6000 * 4000
缩小为1000 * 500
下面为代码实现的主要部分
int is_in_array(short x, short y, short height, short width) { if (x >= 0 && x < width && y >= 0 && y < height) return 1; else return 0; } void bilinera_interpolation(short** in_array, short height, short width, short** out_array, short out_height, short out_width) { double h_times = (double)out_height / (double)height, w_times = (double)out_width / (double)width; short x1, y1, x2, y2, f11, f12, f21, f22; double x, y; for (int i = 0; i < out_height; i++){ for (int j = 0; j < out_width; j++){ x = j / w_times; y = i / h_times; x1 = (short)(x - 1); x2 = (short)(x + 1); y1 = (short)(y + 1); y2 = (short)(y - 1); f11 = is_in_array(x1, y1, height, width) ? in_array[y1][x1] : 0; f12 = is_in_array(x1, y2, height, width) ? in_array[y2][x1] : 0; f21 = is_in_array(x2, y1, height, width) ? in_array[y1][x2] : 0; f22 = is_in_array(x2, y2, height, width) ? in_array[y2][x2] : 0; out_array[i][j] = (short)(((f11 * (x2 - x) * (y2 - y)) + (f21 * (x - x1) * (y2 - y)) + (f12 * (x2 - x) * (y - y1)) + (f22 * (x - x1) * (y - y1))) / ((x2 - x1) * (y2 - y1))); } } }