获取两个连通的点之间所有的坐标点

获取两个连通的点之间所有的坐标点

代码示例

import cv2
import numpy as np

from heapq import (
    heappop,
    heappush,
)


# 定义8连通领域
directions = [(0, 1), (1, 1), (1, 0), (1, -1), (0, -1), (-1, -1), (-1, 0), (-1, 1)]



# 定义一个函数来获取两个点之间连通的所有像素坐标和路径长度
def get_connected_coord_list(mask, point_1, point_2):
    # print(f"==>> point_1: {point_1}")
    # print(f"==>> point_2: {point_2}")
    if len(mask.shape) == 3:
        mask = cv2.cvtColor(mask, cv2.COLOR_BGR2GRAY)

    rows, cols = mask.shape[:2]
    visited = np.zeros_like(mask, dtype=np.uint8)
    queue = []
    # 优先队列，优先级是距离
    # 堆是一种数据结构，它总是保持最小元素在顶部。
    # 将元组(距离, 点)放入堆中，距离较小的点会优先出队
    heappush(queue, (0, point_1))
    # visited[point_1[1], point_1[0]] = 1
    visited[point_1[0], point_1[1]] = 1

    # 记录每个点的父节点，用于回溯路径
    parents = {}
    connected_pixels = []
    while queue:
        # 从对中取出距离最小对应的元组
        current_distance, current_point = heappop(queue)
        if current_point == point_2:
            # print(f"==>> current_point: {current_point}")
            # 
            # # 回溯路径并计算长度
            # path_length = 0
            while current_point in parents:
                # path_length += 1
                current_point = parents[current_point]
                # print(f"==>> current_point: {current_point}")
                connected_pixels.append(current_point)
            # return path_length, connected_pixels
            connected_pixels.append(point_2)
            return connected_pixels

        for direction in directions:
            next_point = (
                current_point[0] + direction[0],
                current_point[1] + direction[1],
            )

            if 0 <= next_point[0] < rows and 0 <= next_point[1] < cols:
                if (
                    visited[next_point[0], next_point[1]] == 0
                    and mask[next_point[0], next_point[1]] > 0
                ):
                    visited[next_point[0], next_point[1]] = 1
                    # 优先队列，优先级是距离
                    heappush(queue, (current_distance + 1, next_point))
                    parents[next_point] = current_point

    # return -1, []
    return []


if __name__ == "__main__":
    mask_path = "./test.png"
    mask = cv2.imread(mask_path, 0)
    # (h, w)
    end_point_crop_list = [(1, 0), (8, 30), (16, 42), (46, 43)]
    intersection_point_crop_list = [(10, 32), (11, 15), (18, 36), (44, 45)]


    # 调用函数并打印结果
    connected_coord_list = get_connected_coord_list(
        mask, (1, 0), (10, 32)
    )
    print(f"==>> connected_coord_list: {connected_coord_list}")

    mask_visual = np.zeros_like(mask)
    for coord in connected_coord_list:
        mask_visual[coord[0], coord[1]] = 255
    print(len(connected_coord_list))
    # 36
    mask_visual_path = "./test_visual_point_mask.png"
    cv2.imwrite(mask_visual_path, mask_visual)