基于触发值检测NumPy向量的上升/下降沿（类示波器触发）

Great question! Traversing arrays with Python loops is definitely slow for large datasets, but NumPy's vectorized operations are perfect for this edge-detection task—they’re implemented in optimized C under the hood, so you’ll get way better performance than manual iteration.

Implementing Rising Edge Detection #

A rising edge is defined as a position where the current value crosses the threshold from below or equal to above. Here's how to implement this efficiently with pure NumPy:

import numpy as np

def rising_edge(data, threshold):
    # Create a mask: current element > threshold AND previous element <= threshold
    edge_mask = (data[1:] > threshold) & (data[:-1] <= threshold)
    # The mask corresponds to gaps between data[0]&data[1], data[1]&data[2], etc.
    # Add 1 to get the index of the rising edge in the original array
    return np.where(edge_mask)[0] + 1

Testing this with your sample data:

data = np.array([-1, -0.5, 0, 0.5, 1, 1.5, 2])
trigger = rising_edge(data, 0.3)
print(trigger)  # Output: [3]

This matches your expected result perfectly.

Implementing Falling Edge Detection #

For falling edges (crossing from above or equal to below the threshold), the logic is nearly identical—just flip the comparisons:

def falling_edge(data, threshold):
    edge_mask = (data[1:] < threshold) & (data[:-1] >= threshold)
    return np.where(edge_mask)[0] + 1

Why This Is Way Faster Than Looping #

NumPy’s vectorized operations avoid the overhead of Python’s loop iteration by processing entire arrays at once in optimized C code. For large datasets (think thousands or millions of points), this will be orders of magnitude faster than a manual loop.

Bonus: Handling Edge Cases & Complex Triggers #

• If you need to ignore tiny "glitches" (short-lived threshold crossings), you can smooth the data first with np.convolve or use morphological filtering from scipy.ndimage to filter out small edges.
• If the first element of your array is already above/below the threshold, it won’t be flagged as an edge—this matches standard oscilloscope trigger behavior, which requires a valid transition.

内容的提问来源于stack exchange，提问作者Cris