最近邻插值、双线性插值及双三次插值方法python

以下是最近邻插值、双线性插值以及双三次插值的Python实现：

最近邻插值：

import numpy as np
from PIL import Image

def nearest_neighbor_interpolation(img, ratio):
    '''
    img: numpy array of shape (height, width, channel)
    ratio: interpolation ratio
    '''
    height, width, channel = img.shape
    new_height = int(height * ratio)
    new_width = int(width * ratio)
    new_img = np.zeros((new_height, new_width, channel), dtype=np.uint8)
    for i in range(new_height):
        for j in range(new_width):
            x = int(i / ratio)
            y = int(j / ratio)
            new_img[i, j, :] = img[x, y, :]
    return new_img

双线性插值：

import numpy as np
from PIL import Image

def bilinear_interpolation(img, ratio):
    '''
    img: numpy array of shape (height, width, channel)
    ratio: interpolation ratio
    '''
    height, width, channel = img.shape
    new_height = int(height * ratio)
    new_width = int(width * ratio)
    new_img = np.zeros((new_height, new_width, channel), dtype=np.uint8)
    for i in range(new_height):
        for j in range(new_width):
            x = i / ratio
            y = j / ratio
            x1 = int(x)
            x2 = x1 + 1
            y1 = int(y)
            y2 = y1 + 1
            if x2 >= height:
                x2 = height - 1
            if y2 >= width:
                y2 = width - 1
            f11 = img[x1, y1, :]
            f12 = img[x1, y2, :]
            f21 = img[x2, y1, :]
            f22 = img[x2, y2, :]
            dx = x - x1
            dy = y - y1
            new_img[i, j, :] = (1 - dx) * (1 - dy) * f11 + dx * (1 - dy) * f21 + (1 - dx) * dy * f12 + dx * dy * f22
    return new_img

双三次插值：

import numpy as np
from PIL import Image

def bicubic_interpolation(img, ratio):
    '''
    img: numpy array of shape (height, width, channel)
    ratio: interpolation ratio
    '''
    def cubic(x, a):
        if abs(x) <= 1:
            return (a + 2) * abs(x) ** 3 - (a + 3) * abs(x) ** 2 + 1
        elif 1 < abs(x) <= 2:
            return a * abs(x) ** 3 - 5 * a * abs(x) ** 2 + 8 * a * abs(x) - 4 * a
        else:
            return 0

    def get_index(x, length):
        if x < 0:
            return 0, 0
        elif x >= length - 1:
            return length - 1, length - 1
        else:
            return int(x), int(x) + 1

    height, width, channel = img.shape
    new_height = int(height * ratio)
    new_width = int(width * ratio)
    new_img = np.zeros((new_height, new_width, channel), dtype=np.uint8)
    for i in range(new_height):
        for j in range(new_width):
            x = i / ratio
            y = j / ratio
            x1, x2 = get_index(x - 1, height)
            x3, x4 = get_index(x, height)
            x5, x6 = get_index(x + 1, height)
            x7, x8 = get_index(x + 2, height)
            y1, y2 = get_index(y - 1, width)
            y3, y4 = get_index(y, width)
            y5, y6 = get_index(y + 1, width)
            y7, y8 = get_index(y + 2, width)
            I1 = cubic((y - y1) / (y2 - y1), -0.5)
            I2 = cubic((y - y3) / (y4 - y3), -0.5)
            I3 = cubic((y - y5) / (y6 - y5), -0.5)
            I4 = cubic((y - y7) / (y8 - y7), -0.5)
            J1 = cubic((x - x1) / (x2 - x1), -0.5)
            J2 = cubic((x - x3) / (x4 - x3), -0.5)
            J3 = cubic((x - x5) / (x6 - x5), -0.5)
            J4 = cubic((x - x7) / (x8 - x7), -0.5)
            f11 = img[x1, y1, :]
            f12 = img[x1, y2, :]
            f13 = img[x1, y3, :]
            f14 = img[x1, y4, :]
            f15 = img[x1, y5, :]
            f16 = img[x1, y6, :]
            f17 = img[x1, y7, :]
            f18 = img[x1, y8, :]
            f21 = img[x2, y1, :]
            f22 = img[x2, y2, :]
            f23 = img[x2, y3, :]
            f24 = img[x2, y4, :]
            f25 = img[x2, y5, :]
            f26 = img[x2, y6, :]
            f27 = img[x2, y7, :]
            f28 = img[x2, y8, :]
            f31 = img[x3, y1, :]
            f32 = img[x3, y2, :]
            f33 = img[x3, y3, :]
            f34 = img[x3, y4, :]
            f35 = img[x3, y5, :]
            f36 = img[x3, y6, :]
            f37 = img[x3, y7, :]
            f38 = img[x3, y8, :]
            f41 = img[x4, y1, :]
            f42 = img[x4, y2, :]
            f43 = img[x4, y3, :]
            f44 = img[x4, y4, :]
            f45 = img[x4, y5, :]
            f46 = img[x4, y6, :]
            f47 = img[x4, y7, :]
            f48 = img[x4, y8, :]
            f51 = img[x5, y1, :]
            f52 = img[x5, y2, :]
            f53 = img[x5, y3, :]
            f54 = img[x5, y4, :]
            f55 = img[x5, y5, :]
            f56 = img[x5, y6, :]
            f57 = img[x5, y7, :]
            f58 = img[x5, y8, :]
            f61 = img[x6, y1, :]
            f62 = img[x6, y2, :]
            f63 = img[x6, y3, :]
            f64 = img[x6, y4, :]
            f65 = img[x6, y5, :]
            f66 = img[x6, y6, :]
            f67 = img[x6, y7, :]
            f68 = img[x6, y8, :]
            f71 = img[x7, y1, :]
            f72 = img[x7, y2, :]
            f73 = img[x7, y3, :]
            f74 = img[x7, y4, :]
            f75 = img[x7, y5, :]
            f76 = img[x7, y6, :]
            f77 = img[x7, y7, :]
            f78 = img[x7, y8, :]
            f81 = img[x8, y1, :]
            f82 = img[x8, y2, :]
            f83 = img[x8, y3, :]
            f84 = img[x8, y4, :]
            f85 = img[x8, y5, :]
            f86 = img[x8, y6, :]
            f87 = img[x8, y7, :]
            f88 = img[x8, y8, :]
            new_img[i, j, :] = (I1 * np.dot(np.array([J1, J2, J3, J4]), np.array([[f11, f12, f13, f14], [f21, f22, f23, f24], [f31, f32, f33, f34], [f41, f42, f43, f44]]))
                                + I2 * np.dot(np.array([J1, J2, J3, J4]), np.array([[f21, f22, f23, f24], [f31, f32, f33, f34], [f41, f42, f43, f44], [f51, f52, f53, f54]]))
                                + I3 * np.dot(np.array([J1, J2, J3, J4]), np.array([[f31, f32, f33, f34], [f41, f42, f43, f44], [f51, f52, f53, f54], [f61, f62, f63, f64]]))
                                + I4 * np.dot(np.array([J1, J2, J3, J4]), np.array([[f41, f42, f43, f44], [f51, f52, f53, f54], [f61, f62, f63, f64], [f71, f72, f73, f74]])))
    return new_img

以上是三种插值方法的Python实现，可以在实际应用中根据需要进行选择。