ValueError: the input array must have size 3 along `channel_axis`, got (417, 556)

这个错误通常是因为在读取图像时，使用的是灰度图像而不是彩色图像，导致图像只有一个通道，而DWT变换所需要的图像必须包含三个通道。

解决方法是将灰度图像转换为RGB图像，并对RGB图像的三个通道进行DWT变换，最后将三个通道的系数合并起来。

下面是修改后的代码：

import numpy as np
import matplotlib.pyplot as plt
import pywt
from skimage import io, color

# 读取灰度图像并转换为RGB图像
img_gray = io.imread('lena.png', as_gray=True)
img = color.gray2rgb(img_gray)

# 对图像的三个通道进行DWT变换
coeffs_r = pywt.dwt2(img[:, :, 0], 'haar')
coeffs_g = pywt.dwt2(img[:, :, 1], 'haar')
coeffs_b = pywt.dwt2(img[:, :, 2], 'haar')

# 设置压缩比率
compress_ratio = 0.5

# 计算阈值
threshold_r = np.sort(np.abs(coeffs_r[1].ravel()))[::-1][int(compress_ratio * len(coeffs_r[1].ravel()))]
threshold_g = np.sort(np.abs(coeffs_g[1].ravel()))[::-1][int(compress_ratio * len(coeffs_g[1].ravel()))]
threshold_b = np.sort(np.abs(coeffs_b[1].ravel()))[::-1][int(compress_ratio * len(coeffs_b[1].ravel()))]

# 对小于阈值的系数进行置零
coeffs_r = list(coeffs_r)
coeffs_r[0] = np.round(coeffs_r[0])
coeffs_r[1] = np.where(np.abs(coeffs_r[1]) < threshold_r, 0, coeffs_r[1])
coeffs_r[2] = np.where(np.abs(coeffs_r[2]) < threshold_r, 0, coeffs_r[2])
coeffs_g = list(coeffs_g)
coeffs_g[0] = np.round(coeffs_g[0])
coeffs_g[1] = np.where(np.abs(coeffs_g[1]) < threshold_g, 0, coeffs_g[1])
coeffs_g[2] = np.where(np.abs(coeffs_g[2]) < threshold_g, 0, coeffs_g[2])
coeffs_b = list(coeffs_b)
coeffs_b[0] = np.round(coeffs_b[0])
coeffs_b[1] = np.where(np.abs(coeffs_b[1]) < threshold_b, 0, coeffs_b[1])
coeffs_b[2] = np.where(np.abs(coeffs_b[2]) < threshold_b, 0, coeffs_b[2])

# 合并三个通道的系数
coeffs = [np.stack([coeffs_r[i], coeffs_g[i], coeffs_b[i]], axis=-1) for i in range(len(coeffs_r))]

# 对图像进行IDWT反变换
img_dwt = pywt.idwt2(coeffs, 'haar')

# 显示原始图像和压缩后的图像
fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(8, 4))
ax = axes.ravel()

ax[0].imshow(img)
ax[0].set_title("Original image")

ax[1].imshow(img_dwt)
ax[1].set_title("Compressed image (DWT)")
ax[1].set_xlabel("Compression ratio: {:.2f}".format(compress_ratio))

plt.tight_layout()
plt.show()

在这个修改后的代码中，我们首先使用`as_gray=True`将图像读取为灰度图像，然后使用`color.gray2rgb`将灰度图像转换为RGB图像。然后对RGB图像的三个通道分别进行DWT变换，并计算出三个通道的阈值，最后将三个通道的系数合并起来，并进行IDWT反变换。