垂直线条聚类分组问题求助：现有算法存在误差需优化

我需要将坐标（x1,y1,x2,y2）相近的线条进行分组，已成功实现水平线分组，但垂直线分组存在误差。我采用了类似水平线的处理方法（仅反向排序并匹配Y轴操作），但结果未达预期。图像尺寸为2675*2275，恳请大家提供帮助。

问题代码 #

def connect_close_vertical_lines(Test_line_bg, group, distance_threshold=30):

    sorted_group = sorted(group, key=lambda line: line[1])
    # sorted_group_vertical_avg = []
    # total_x = 0
    # len_x = 0
    # for line in sorted_group:
    #     x1, y1, x2, y2 = line[:4]
    #     total_x += x1
    #     len_x += 1
    # avg_x = total_x / len_x
    # for line in sorted_group:
    #     x1, y1, x2, y2 = line[:4]
    #     x1 = x2 = avg_x
    #     sorted_group_vertical_avg.append((x1, y1, x2, y2))

    connected_line = sorted_group[0]
    for i in range(1, len(sorted_group)):
        line = sorted_group[i]
        x1, y1, x2, y2 = map(int, connected_line[:4])
        x3, y3, x4, y4 = map(int, line[:4])


        dist_2 = abs(x2 - x3)
        if dist_2 < 30:
            distance_threshold = distance_threshold*2
        dist = abs(y2 - y3)


        if dist < distance_threshold:
            connected_line = (x1, y1, x1, y4)

        else:
            connected_lines_drawn_vertical.append(connected_line)

            cv2.line(Test_line_bg, (x1, y1), (x1, y2), (255, 255, 255), 3)


            connected_line = line
    connected_lines_drawn_vertical.append(connected_line)

    x1, y1, x2, y2 = map(int, connected_line[:4])
    cv2.line(Test_line_bg, (x1, y1), (x1, y2), (255, 255, 255), 3)

补充说明 #

• 效果示意图：显示部分垂直线未被正确合并，仍存在零散短线条
• 坐标数据：包含待处理的垂直线坐标集合

内容的提问来源于stack exchange，提问作者lee