如何使用matplotlib FuncAnimation实现numpy数组滑动窗口的动画展示

核心逻辑 #

• frames参数直接传入所有滑动窗口的起始偏移量，范围为0到len(信号数组)-窗口长度，每个frame值对应当前窗口的左边界
• 初始化函数仅创建空的线条、散点对象，固定y轴范围避免画面跳动
• 逐帧更新函数接收frame作为参数，按窗口范围切片信号数据，筛选出落在该区间内的峰值，更新绘图对象的数据即可

效果示例 #

完整可运行代码 #

import neurokit2 as nk
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
from scipy.signal import find_peaks
from matplotlib.animation import FuncAnimation

# 配置参数
WINDOW_SIZE = 500  # 5秒窗口，对应100Hz采样率
SAMPLING_RATE = 100

# 生成模拟数据、检测峰值
data = nk.ecg_simulate(duration = 50, sampling_rate = SAMPLING_RATE, noise = 0.05, random_state = 1)
scaler = MinMaxScaler()
scaled_arr = scaler.fit_transform(data.reshape(-1,1)).squeeze()
peak_x, peak_info = find_peaks(scaled_arr, height = .66, distance = 60, prominence = .5)
peak_y = peak_info['peak_heights']

# 初始化画布
fig, ax = plt.subplots(figsize=(10,4))
# 初始化空绘图对象
line, = ax.plot([], [], lw=1)
scatter = ax.scatter([], [], s=20, c='red')
# 固定轴范围
ax.set_ylim(scaled_arr.min()-0.05, scaled_arr.max()+0.05)
ax.set_xlim(0, WINDOW_SIZE)
ax.set_xlabel('窗口内时间点')
ax.set_ylabel('归一化信号值')

def init():
    line.set_data([], [])
    scatter.set_offsets([])
    return line, scatter

def update(frame):
    # 切片当前窗口的信号
    window_signal = scaled_arr[frame : frame+WINDOW_SIZE]
    line.set_data(range(WINDOW_SIZE), window_signal)
    # 筛选当前窗口内的峰值
    mask = (peak_x >= frame) & (peak_x < frame + WINDOW_SIZE)
    current_peak_x = peak_x[mask] - frame  # 转为窗口内相对坐标
    current_peak_y = peak_y[mask]
    scatter.set_offsets(list(zip(current_peak_x, current_peak_y)))
    return line, scatter

# 生成动画
ani = FuncAnimation(
    fig,
    update,
    frames=range(0, len(scaled_arr)-WINDOW_SIZE),
    init_func=init,
    interval=20,  # 每帧间隔20ms，对应1倍速滑动
    blit=True,
    repeat=False
)

plt.show()
# 如需保存为视频可取消下方注释
# ani.save('peak_demo.mp4', writer='ffmpeg', fps=50)

内容的提问来源于stack exchange，提问作者fednem