Python Matplotlib背景 colored颜色渐变生成边界

发布于02月07日

我希望我的绘图的白色背景与渐变混合在一起，并让该 colored颜色渐变填充从某个x值开始的绘图的右半部分. 我通过将 colored颜色渐变插入到matplotlib绘图的背景中，使用imshow并使用海运调色板来实现这一点.

然而，如果我使用下面的代码这样做，我会得到调色板中最浅的 colored颜色 (我希望它是白色)和绘图背景的白色之间的轻微不匹配.

 ax.imshow(
    [[0, 1], [0, 1]],
    interpolation="bicubic",
    cmap=sns.cubehelix_palette(start=0, rot=0, hue=0, light=1, dark=0.7, as_cmap=True,),
    extent=(x_min, x_max, y_min, y_max),
    alpha=1,
    aspect="auto",
)

问题与以下情况相同

cmap=sns.color_palette("blend:#FFFFFF,#C1C1C1", as_cmap=True)

The difference is subtle but noticable as seen from the attached image. How can I resolve this and match the two colors?

推荐答案

当只给出两个值时，interpolation="bicubic"会进行一些外推.相反，您可以使用interpolation="bilinear".

要获得三次曲线的插值值，可以将[[0, 1], [0, 1]]更改为三次函数.

以下是提取每种情况的像素值并将其显示为曲线的代码.

import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np

fig, axs = plt.subplots(ncols=3, nrows=2, sharey='row', figsize=(15, 6))
for i, interp in enumerate(['bicubic', 'bilinear', 'bilinear']):
    if i < 2:
        gradient = [[0, 1], [0, 1]]
    else:
        x = np.linspace(0, 1, 100)
        gradient = ((3 - 2 * x) * x * x).reshape(1, -1)
    img = axs[0, i].imshow(gradient,
                           interpolation=interp,
                           cmap=sns.cubehelix_palette(start=0, rot=0, hue=0, light=1, dark=0.7, as_cmap=True),
                           extent=[0, 1, 0, 1],
                           alpha=1, aspect="auto")
    axs[0, i].set_title(f'interpolation={interp} {"linear" if i < 2 else "cubic"} gradient')
    axs[0, i].set_xticks([])
    axs[0, i].set_yticks([])

    # show the color value of the central row as a curve
    data = img.make_image(renderer=None)[0]  # get image data as a numpy NxMx4 array
    values = data[data.shape[0] // 2, :, :3].mean(axis=1)  # get central row and average r, g and b
    axs[1, i].step(np.arange(len(values)), values, color='dodgerblue')
    axs[1, i].axhline(255, color='red', ls='--')
    axs[1, i].text(len(values), 255, "\n255 (100% white) ", color='red', va='center', ha='right')
    axs[1, i].set_title(f'start pixel: {values[0]:.0f}')
    axs[1, i].margins(x=0)
axs[1, 0].set_ylabel('color values')

plt.tight_layout()
plt.show()