Python提取Mesh边界边生成多边形：仅获小三角边界的问题

核心问题 #

1. 未筛选边界边：你当前把所有面的边都加入了列表，包括内部边（每个内部边会在两个相邻面中各出现一次），但边界边的定义是仅在单个面中出现的边，这是导致你得到错误结果的根本原因。
2. 用坐标存储边存在精度风险：直接用浮点坐标作为边的标识，可能因浮点精度误差导致同一个边被判定为不同的边，影响后续的去重和遍历逻辑。
3. 边的遍历逻辑依赖错误的边集合：基于包含内部边的集合构建多边形，自然无法得到正确的外边界。

修复后的代码 #

import plyfile

def get_boundary_edges(faces):
    # 用顶点索引统计每条边的出现次数，边用排序后的索引元组存储（确保(v1,v2)和(v2,v1)视为同一条边）
    edge_count = {}
    for face in faces:
        vertex_indices = face['vertex_indices']
        face_len = len(vertex_indices)
        for i in range(face_len):
            v1 = vertex_indices[i]
            v2 = vertex_indices[(i+1) % face_len]
            # 排序索引，统一边的存储形式
            edge = tuple(sorted((v1, v2)))
            edge_count[edge] = edge_count.get(edge, 0) + 1
    
    # 筛选出仅出现一次的边（边界边），并恢复为有向边的形式
    boundary_edges = []
    for face in faces:
        vertex_indices = face['vertex_indices']
        face_len = len(vertex_indices)
        for i in range(face_len):
            v1 = vertex_indices[i]
            v2 = vertex_indices[(i+1) % face_len]
            edge = tuple(sorted((v1, v2)))
            if edge_count[edge] == 1:
                boundary_edges.append((v1, v2))
    
    return boundary_edges

def edges_to_polygon_loop(boundary_edges, vertices):
    if not boundary_edges:
        return []
    
    # 构建顶点到相邻边的映射，方便快速查找下一条边
    vertex_edge_map = {}
    for edge in boundary_edges:
        v_start, v_end = edge
        if v_start not in vertex_edge_map:
            vertex_edge_map[v_start] = []
        vertex_edge_map[v_start].append(edge)
        if v_end not in vertex_edge_map:
            vertex_edge_map[v_end] = []
        vertex_edge_map[v_end].append(edge)
    
    # 从第一条边开始构建环
    start_edge = boundary_edges[0]
    current_v = start_edge[1]
    polygon_loop = [vertices[start_edge[0]], vertices[current_v]]
    visited_edges = set([start_edge])
    
    while current_v != start_edge[0]:
        # 找到当前顶点出发的未访问边
        for edge in vertex_edge_map.get(current_v, []):
            if edge not in visited_edges:
                next_v = edge[1] if edge[0] == current_v else edge[0]
                polygon_loop.append(vertices[next_v])
                visited_edges.add(edge)
                current_v = next_v
                break
    
    return polygon_loop

# 读取PLY文件
with open('M:\\volume.ply', 'rb') as f:
    ply = plyfile.PlyData.read(f)
    vertices = ply['vertex']
    faces = ply['face']

# 获取边界边
boundary_edges = get_boundary_edges(faces)

# 构建多边形环
polygon = edges_to_polygon_loop(boundary_edges, vertices)

# 输出结果示例
print("生成的多边形顶点数量:", len(polygon))

关键修复点说明 #

• 筛选边界边：通过统计每条边的出现次数，仅保留出现一次的边，确保后续处理的是真正的边界。
• 用顶点索引处理边：避免浮点坐标的精度问题，同时统一边的存储形式（排序索引），确保(v1,v2)和(v2,v1)被识别为同一条边。
• 优化遍历逻辑：构建顶点到边的映射，提升查找下一条边的效率，同时保证环的正确闭合。

内容的提问来源于stack exchange，提问作者Purple_Ad