如何对Numpy二维数组进行三角形区域切片？

Got it, let's break down how to solve this problem step by step. First, let's clarify the structure of your target triangle: the three points (1,1), (3,1), (3,5) (0-based indices, row first, column second) form a right triangle with the right angle at (3,1). From your expected result, we can see each row's valid column range starts at column 1 and ends at 2*y -1 (where y is the current row index).

Here are two implementation approaches for different scenarios:

方法一：逐行赋值（简单直观，适合小数组） #

This method is straightforward and easy to understand, perfect for quick validation of your需求:

import numpy as np

# Create the original 4x8 zero array
arr = np.zeros((4, 8), dtype=int)

# Iterate over rows covered by the triangle (row indices 1 to 3)
for y in range(1, 4):
    # Calculate the rightmost column index for current row
    x_right = 2 * y - 1
    # Assign 1 to the range [1, x_right] (slice is left-closed right-open, so +1 is needed)
    arr[y, 1 : x_right + 1] = 1

# Print the result
print(arr)

Running this code will output exactly your expected result:

[[0 0 0 0 0 0 0 0]
 [0 1 0 0 0 0 0 0]
 [0 1 1 1 0 0 0 0]
 [0 1 1 1 1 1 0 0]]

方法二：向量化掩码（高效无循环，适合大数组） #

If you're working with large arrays, loops can be slow. Using Numpy's vectorized operations is a better choice. We'll generate grid coordinates, create a mask based on triangle conditions, then assign values in bulk:

import numpy as np

arr = np.zeros((4, 8), dtype=int)

# Generate grid coordinate matrices, indexing='ij' ensures row-first order
y_coords, x_coords = np.meshgrid(np.arange(4), np.arange(8), indexing='ij')

# Build the mask: conditions for points inside the triangle
mask = (y_coords >= 1) &       # Rows are between 1 and 3
       (y_coords <= 3) &       #
       (x_coords >= 1) &       # Columns start from 1
       (x_coords <= 2*y_coords - 1)  # Columns don't exceed the hypotenuse boundary

# Assign 1 to all positions covered by the mask
arr[mask] = 1

print(arr)

This approach avoids loops entirely, leveraging Numpy's broadcasting for batch operations, which is much more efficient for large-scale data.

Key Logic Explanation #

The hypotenuse equation is derived from the two points (1,1) and (3,5): the slope is (5-1)/(3-1)=2, so the line equation is x = 2*(y-1) + 1 = 2y -1. This gives the maximum column index for each row y, ensuring we exactly cover the target triangle area.

内容的提问来源于stack exchange，提问作者Ali