从离散点云到三角网用Python/C手把手实现Delaunay三角剖分附完整代码在三维建模、地理信息系统和游戏开发中将离散的点云数据转换为连续的三角网格是一项基础而关键的技术。Delaunay三角剖分因其优秀的数学特性——最大化最小角、避免狭长三角形——成为最常用的三角化方法之一。本文将深入解析Delaunay三角网的生长算法原理并分别用Python和C实现完整的三角剖分流程最后讨论如何将生成的网格导出为OBJ等通用格式。1. Delaunay三角剖分核心原理Delaunay三角剖分的核心是空圆特性对于任意一个三角形其外接圆内不包含其他点。这一特性保证了三角网的质量避免了银三角形的出现。实现这一特性的常用判断方法包括余弦判断法计算基线两端点到候选点的向量夹角余弦选择余弦值最小的点叉积判断法通过向量叉积符号判断点是否在基线的正确侧空圆判断直接计算外接圆是否包含其他点计算量较大数学上给定基线P1P2和候选点P余弦值计算公式为cosθ (P1P · P2P) / (|P1P| * |P2P|)其中P1P和P2P分别表示从P1到P和P2到P的向量。最小余弦值对应最大夹角最符合Delaunay准则。2. 算法实现框架设计2.1 基础数据结构无论是Python还是C实现都需要定义几个核心类# Python版本基础类 class Point: def __init__(self, x, y, z0): self.x x self.y y self.z z def distance_to(self, other): return ((self.x-other.x)**2 (self.y-other.y)**2)**0.5 class Triangle: def __init__(self, p1, p2, p3): self.vertices [p1, p2, p3] self.edges [(p1,p2), (p2,p3), (p3,p1)]// C版本基础类 class Point { public: double x, y, z; Point(double x0, double y0, double z0) : x(x), y(y), z(z) {} double distanceTo(const Point other) const { return sqrt(pow(x-other.x, 2) pow(y-other.y, 2)); } }; class Triangle { public: Point vertices[3]; std::pairPoint, Point edges[3]; Triangle(Point p1, Point p2, Point p3) { vertices[0] p1; vertices[1] p2; vertices[2] p3; edges[0] {p1,p2}; edges[1] {p2,p3}; edges[2] {p3,p1}; } };2.2 生长算法流程完整的Delaunay生长算法可分为五个阶段初始化输入离散点集S找到距离最近的两点P1,P2作为初始基线构建首三角形在基线右侧寻找使余弦值最小的点P3形成初始三角形T(P1,P2,P3)边扩展将首三角形的三条边加入待处理边队列对每条边在其相反侧寻找满足空圆准则的点迭代处理从队列取出一条边寻找最佳扩展点形成新三角形将新边加入队列更新边的使用计数终止条件所有边都已被处理使用计数≥2剩余点无法形成有效三角形3. Python完整实现以下是基于生长算法的Python实现包含完整的Delaunay三角剖分import math import random from collections import defaultdict class DelaunayTriangulation: def __init__(self, points): self.points points self.triangles [] self.edge_usage defaultdict(int) def find_closest_pair(self): min_dist float(inf) pair None for i in range(len(self.points)): for j in range(i1, len(self.points)): dist self.points[i].distance_to(self.points[j]) if dist min_dist: min_dist dist pair (self.points[i], self.points[j]) return pair def find_third_point(self, p1, p2, points): min_cos float(inf) best_point None for p in points: if p p1 or p p2: continue # 计算向量 v1 (p1.x - p.x, p1.y - p.y) v2 (p2.x - p.x, p2.y - p.y) # 计算点积和模 dot v1[0]*v2[0] v1[1]*v2[1] mod math.sqrt(v1[0]**2 v1[1]**2) * math.sqrt(v2[0]**2 v2[1]**2) # 计算余弦值 cos_val dot / mod if cos_val min_cos: min_cos cos_val best_point p return best_point def edge_key(self, p1, p2): return tuple(sorted([(p1.x, p1.y), (p2.x, p2.y)])) def triangulate(self): if len(self.points) 3: return [] # 步骤1找到最近点对 p1, p2 self.find_closest_pair() remaining_points [p for p in self.points if p ! p1 and p ! p2] # 步骤2构建首三角形 p3 self.find_third_point(p1, p2, remaining_points) if not p3: return [] initial_triangle Triangle(p1, p2, p3) self.triangles.append(initial_triangle) # 初始化边队列 edge_queue [] for edge in initial_triangle.edges: edge_key self.edge_key(edge[0], edge[1]) self.edge_usage[edge_key] 1 edge_queue.append(edge) # 步骤3迭代扩展 while edge_queue: current_edge edge_queue.pop(0) edge_key self.edge_key(current_edge[0], current_edge[1]) if self.edge_usage[edge_key] 2: continue # 寻找扩展点 opposite_points [] for triangle in self.triangles: if current_edge in triangle.edges: for vertex in triangle.vertices: if vertex not in current_edge: opposite_point vertex break # 在另一侧寻找候选点 candidates [] a, b current_edge for p in remaining_points: # 使用叉积判断是否在另一侧 cross (b.x - a.x)*(p.y - a.y) - (b.y - a.y)*(p.x - a.x) if cross * ((b.x - a.x)*(opposite_point.y - a.y) - (b.y - a.y)*(opposite_point.x - a.x)) 0: candidates.append(p) if not candidates: continue # 选择最佳候选点 best_point self.find_third_point(current_edge[0], current_edge[1], candidates) if not best_point: continue # 创建新三角形 new_triangle Triangle(current_edge[0], current_edge[1], best_point) self.triangles.append(new_triangle) # 更新边状态 for edge in new_triangle.edges: e_key self.edge_key(edge[0], edge[1]) if e_key edge_key: self.edge_usage[e_key] 1 else: if e_key in self.edge_usage: self.edge_usage[e_key] 1 else: self.edge_usage[e_key] 1 edge_queue.append(edge) return self.triangles4. C优化实现C版本在性能上更有优势适合处理大规模点云#include vector #include cmath #include map #include algorithm class DelaunayCpp { public: struct Edge { Point p1, p2; bool operator(const Edge other) const { auto make_tuple [](const Point p) { return std::tie(p.x, p.y); }; auto t1 std::minmax(make_tuple(p1), make_tuple(p2)); auto t2 std::minmax(make_tuple(other.p1), make_tuple(other.p2)); return t1 t2; } }; std::vectorTriangle triangulate(const std::vectorPoint points) { std::vectorTriangle triangles; if (points.size() 3) return triangles; // 复制点集用于处理 std::vectorPoint remaining_points points; // 步骤1找到最近点对 auto closest_pair findClosestPair(remaining_points); Point p1 closest_pair.first, p2 closest_pair.second; // 从剩余点中移除这两个点 remaining_points.erase(std::remove(remaining_points.begin(), remaining_points.end(), p1), remaining_points.end()); remaining_points.erase(std::remove(remaining_points.begin(), remaining_points.end(), p2), remaining_points.end()); // 步骤2构建首三角形 Point p3 findThirdPoint(p1, p2, remaining_points); if (p3 Point()) return triangles; triangles.emplace_back(p1, p2, p3); // 初始化边队列 std::mapEdge, int edge_usage; std::vectorEdge edge_queue; auto add_edge [](const Point a, const Point b) { Edge e{a, b}; edge_usage[e] 1; edge_queue.push_back(e); }; add_edge(p1, p2); add_edge(p2, p3); add_edge(p3, p1); // 步骤3迭代扩展 while (!edge_queue.empty()) { Edge current_edge edge_queue.front(); edge_queue.erase(edge_queue.begin()); if (edge_usage[current_edge] 2) continue; // 找到当前边所属的三角形和对面点 Point opposite_point; for (const auto tri : triangles) { bool has_edge false; for (const auto edge : tri.edges) { if ((edge.first current_edge.p1 edge.second current_edge.p2) || (edge.first current_edge.p2 edge.second current_edge.p1)) { has_edge true; break; } } if (has_edge) { for (const auto v : tri.vertices) { if (v ! current_edge.p1 v ! current_edge.p2) { opposite_point v; break; } } break; } } // 在另一侧寻找候选点 std::vectorPoint candidates; for (const auto p : remaining_points) { double cross_current crossProduct(current_edge.p1, current_edge.p2, opposite_point); double cross_candidate crossProduct(current_edge.p1, current_edge.p2, p); if (cross_current * cross_candidate 0) { candidates.push_back(p); } } if (candidates.empty()) continue; // 选择最佳候选点 Point best_point findThirdPoint(current_edge.p1, current_edge.p2, candidates); if (best_point Point()) continue; // 创建新三角形 triangles.emplace_back(current_edge.p1, current_edge.p2, best_point); // 更新边状态 auto update_edge [](const Point a, const Point b) { Edge e{a, b}; if (e current_edge || current_edge e) { // 不是当前边 edge_usage[e]; edge_queue.push_back(e); } else { edge_usage[current_edge]; } }; update_edge(current_edge.p1, best_point); update_edge(current_edge.p2, best_point); } return triangles; } private: std::pairPoint, Point findClosestPair(std::vectorPoint points) { double min_dist std::numeric_limitsdouble::max(); std::pairPoint, Point closest_pair; for (size_t i 0; i points.size(); i) { for (size_t j i 1; j points.size(); j) { double dist points[i].distanceTo(points[j]); if (dist min_dist) { min_dist dist; closest_pair {points[i], points[j]}; } } } return closest_pair; } Point findThirdPoint(const Point p1, const Point p2, const std::vectorPoint candidates) { if (candidates.empty()) return Point(); double min_cos std::numeric_limitsdouble::max(); Point best_point; for (const auto p : candidates) { // 计算向量 double v1x p1.x - p.x, v1y p1.y - p.y; double v2x p2.x - p.x, v2y p2.y - p.y; // 计算点积和模 double dot v1x * v2x v1y * v2y; double mod std::sqrt(v1x*v1x v1y*v1y) * std::sqrt(v2x*v2x v2y*v2y); // 避免除零错误 if (mod 1e-10) continue; double cos_val dot / mod; if (cos_val min_cos) { min_cos cos_val; best_point p; } } return best_point; } double crossProduct(const Point a, const Point b, const Point p) { return (b.x - a.x) * (p.y - a.y) - (b.y - a.y) * (p.x - a.x); } };5. 性能优化与格式导出5.1 算法优化技巧空间索引加速使用网格或四叉树空间索引来加速最近邻搜索增量式更新维护一个活动边表避免重复计算并行计算对独立区域可并行处理# 使用网格空间索引的Python示例 class SpatialGrid: def __init__(self, points, cell_size): self.cell_size cell_size self.grid defaultdict(list) min_x min(p.x for p in points) min_y min(p.y for p in points) for p in points: cell_x int((p.x - min_x) / cell_size) cell_y int((p.y - min_y) / cell_size) self.grid[(cell_x, cell_y)].append(p) def query_neighbors(self, point, radius): min_x point.x - radius min_y point.y - radius max_x point.x radius max_y point.y radius candidates [] for cell_x in range(int(min_x/self.cell_size), int(max_x/self.cell_size)1): for cell_y in range(int(min_y/self.cell_size), int(max_y/self.cell_size)1): candidates.extend(self.grid.get((cell_x, cell_y), [])) return [p for p in candidates if min_x p.x max_x and min_y p.y max_y]5.2 导出为OBJ格式OBJ是一种简单的3D模型格式适合三角网格导出def export_to_obj(triangles, filename): with open(filename, w) as f: # 收集所有顶点 vertices [] vertex_index {} for tri in triangles: for v in tri.vertices: if (v.x, v.y, v.z) not in vertex_index: vertex_index[(v.x, v.y, v.z)] len(vertices) 1 vertices.append(v) # 写入顶点 for v in vertices: f.write(fv {v.x} {v.y} {v.z}\n) # 写入面 for tri in triangles: indices [vertex_index[(v.x, v.y, v.z)] for v in tri.vertices] f.write(ff {indices[0]} {indices[1]} {indices[2]}\n)C版本的OBJ导出void exportToObj(const std::vectorTriangle triangles, const std::string filename) { std::ofstream file(filename); if (!file.is_open()) return; std::mapstd::tupledouble, double, double, int vertex_index; std::vectorPoint vertices; // 收集顶点并建立索引 for (const auto tri : triangles) { for (const auto v : tri.vertices) { auto key std::make_tuple(v.x, v.y, v.z); if (vertex_index.find(key) vertex_index.end()) { vertex_index[key] vertices.size() 1; vertices.push_back(v); } } } // 写入顶点 for (const auto v : vertices) { file v v.x v.y v.z \n; } // 写入面 for (const auto tri : triangles) { file f ; for (const auto v : tri.vertices) { auto key std::make_tuple(v.x, v.y, v.z); file vertex_index[key] ; } file \n; } file.close(); }6. 两种语言实现对比特性Python实现C实现开发效率高代码简洁较低需要更多样板代码运行性能较慢适合小规模数据快适合大规模点云处理内存管理自动垃圾回收手动/智能指针管理多线程支持受GIL限制完整的多线程支持第三方库生态丰富NumPy, SciPy等较少但性能优化库多适合场景原型开发、小数据处理生产环境、高性能需求在实际项目中可以根据需求选择实现方式。对于需要快速验证算法或处理中小规模数据Python是更好的选择而对于性能要求高的生产环境C实现更为合适。
从离散点云到三角网：用Python/C++手把手实现Delaunay三角剖分（附完整代码）
从离散点云到三角网用Python/C手把手实现Delaunay三角剖分附完整代码在三维建模、地理信息系统和游戏开发中将离散的点云数据转换为连续的三角网格是一项基础而关键的技术。Delaunay三角剖分因其优秀的数学特性——最大化最小角、避免狭长三角形——成为最常用的三角化方法之一。本文将深入解析Delaunay三角网的生长算法原理并分别用Python和C实现完整的三角剖分流程最后讨论如何将生成的网格导出为OBJ等通用格式。1. Delaunay三角剖分核心原理Delaunay三角剖分的核心是空圆特性对于任意一个三角形其外接圆内不包含其他点。这一特性保证了三角网的质量避免了银三角形的出现。实现这一特性的常用判断方法包括余弦判断法计算基线两端点到候选点的向量夹角余弦选择余弦值最小的点叉积判断法通过向量叉积符号判断点是否在基线的正确侧空圆判断直接计算外接圆是否包含其他点计算量较大数学上给定基线P1P2和候选点P余弦值计算公式为cosθ (P1P · P2P) / (|P1P| * |P2P|)其中P1P和P2P分别表示从P1到P和P2到P的向量。最小余弦值对应最大夹角最符合Delaunay准则。2. 算法实现框架设计2.1 基础数据结构无论是Python还是C实现都需要定义几个核心类# Python版本基础类 class Point: def __init__(self, x, y, z0): self.x x self.y y self.z z def distance_to(self, other): return ((self.x-other.x)**2 (self.y-other.y)**2)**0.5 class Triangle: def __init__(self, p1, p2, p3): self.vertices [p1, p2, p3] self.edges [(p1,p2), (p2,p3), (p3,p1)]// C版本基础类 class Point { public: double x, y, z; Point(double x0, double y0, double z0) : x(x), y(y), z(z) {} double distanceTo(const Point other) const { return sqrt(pow(x-other.x, 2) pow(y-other.y, 2)); } }; class Triangle { public: Point vertices[3]; std::pairPoint, Point edges[3]; Triangle(Point p1, Point p2, Point p3) { vertices[0] p1; vertices[1] p2; vertices[2] p3; edges[0] {p1,p2}; edges[1] {p2,p3}; edges[2] {p3,p1}; } };2.2 生长算法流程完整的Delaunay生长算法可分为五个阶段初始化输入离散点集S找到距离最近的两点P1,P2作为初始基线构建首三角形在基线右侧寻找使余弦值最小的点P3形成初始三角形T(P1,P2,P3)边扩展将首三角形的三条边加入待处理边队列对每条边在其相反侧寻找满足空圆准则的点迭代处理从队列取出一条边寻找最佳扩展点形成新三角形将新边加入队列更新边的使用计数终止条件所有边都已被处理使用计数≥2剩余点无法形成有效三角形3. Python完整实现以下是基于生长算法的Python实现包含完整的Delaunay三角剖分import math import random from collections import defaultdict class DelaunayTriangulation: def __init__(self, points): self.points points self.triangles [] self.edge_usage defaultdict(int) def find_closest_pair(self): min_dist float(inf) pair None for i in range(len(self.points)): for j in range(i1, len(self.points)): dist self.points[i].distance_to(self.points[j]) if dist min_dist: min_dist dist pair (self.points[i], self.points[j]) return pair def find_third_point(self, p1, p2, points): min_cos float(inf) best_point None for p in points: if p p1 or p p2: continue # 计算向量 v1 (p1.x - p.x, p1.y - p.y) v2 (p2.x - p.x, p2.y - p.y) # 计算点积和模 dot v1[0]*v2[0] v1[1]*v2[1] mod math.sqrt(v1[0]**2 v1[1]**2) * math.sqrt(v2[0]**2 v2[1]**2) # 计算余弦值 cos_val dot / mod if cos_val min_cos: min_cos cos_val best_point p return best_point def edge_key(self, p1, p2): return tuple(sorted([(p1.x, p1.y), (p2.x, p2.y)])) def triangulate(self): if len(self.points) 3: return [] # 步骤1找到最近点对 p1, p2 self.find_closest_pair() remaining_points [p for p in self.points if p ! p1 and p ! p2] # 步骤2构建首三角形 p3 self.find_third_point(p1, p2, remaining_points) if not p3: return [] initial_triangle Triangle(p1, p2, p3) self.triangles.append(initial_triangle) # 初始化边队列 edge_queue [] for edge in initial_triangle.edges: edge_key self.edge_key(edge[0], edge[1]) self.edge_usage[edge_key] 1 edge_queue.append(edge) # 步骤3迭代扩展 while edge_queue: current_edge edge_queue.pop(0) edge_key self.edge_key(current_edge[0], current_edge[1]) if self.edge_usage[edge_key] 2: continue # 寻找扩展点 opposite_points [] for triangle in self.triangles: if current_edge in triangle.edges: for vertex in triangle.vertices: if vertex not in current_edge: opposite_point vertex break # 在另一侧寻找候选点 candidates [] a, b current_edge for p in remaining_points: # 使用叉积判断是否在另一侧 cross (b.x - a.x)*(p.y - a.y) - (b.y - a.y)*(p.x - a.x) if cross * ((b.x - a.x)*(opposite_point.y - a.y) - (b.y - a.y)*(opposite_point.x - a.x)) 0: candidates.append(p) if not candidates: continue # 选择最佳候选点 best_point self.find_third_point(current_edge[0], current_edge[1], candidates) if not best_point: continue # 创建新三角形 new_triangle Triangle(current_edge[0], current_edge[1], best_point) self.triangles.append(new_triangle) # 更新边状态 for edge in new_triangle.edges: e_key self.edge_key(edge[0], edge[1]) if e_key edge_key: self.edge_usage[e_key] 1 else: if e_key in self.edge_usage: self.edge_usage[e_key] 1 else: self.edge_usage[e_key] 1 edge_queue.append(edge) return self.triangles4. C优化实现C版本在性能上更有优势适合处理大规模点云#include vector #include cmath #include map #include algorithm class DelaunayCpp { public: struct Edge { Point p1, p2; bool operator(const Edge other) const { auto make_tuple [](const Point p) { return std::tie(p.x, p.y); }; auto t1 std::minmax(make_tuple(p1), make_tuple(p2)); auto t2 std::minmax(make_tuple(other.p1), make_tuple(other.p2)); return t1 t2; } }; std::vectorTriangle triangulate(const std::vectorPoint points) { std::vectorTriangle triangles; if (points.size() 3) return triangles; // 复制点集用于处理 std::vectorPoint remaining_points points; // 步骤1找到最近点对 auto closest_pair findClosestPair(remaining_points); Point p1 closest_pair.first, p2 closest_pair.second; // 从剩余点中移除这两个点 remaining_points.erase(std::remove(remaining_points.begin(), remaining_points.end(), p1), remaining_points.end()); remaining_points.erase(std::remove(remaining_points.begin(), remaining_points.end(), p2), remaining_points.end()); // 步骤2构建首三角形 Point p3 findThirdPoint(p1, p2, remaining_points); if (p3 Point()) return triangles; triangles.emplace_back(p1, p2, p3); // 初始化边队列 std::mapEdge, int edge_usage; std::vectorEdge edge_queue; auto add_edge [](const Point a, const Point b) { Edge e{a, b}; edge_usage[e] 1; edge_queue.push_back(e); }; add_edge(p1, p2); add_edge(p2, p3); add_edge(p3, p1); // 步骤3迭代扩展 while (!edge_queue.empty()) { Edge current_edge edge_queue.front(); edge_queue.erase(edge_queue.begin()); if (edge_usage[current_edge] 2) continue; // 找到当前边所属的三角形和对面点 Point opposite_point; for (const auto tri : triangles) { bool has_edge false; for (const auto edge : tri.edges) { if ((edge.first current_edge.p1 edge.second current_edge.p2) || (edge.first current_edge.p2 edge.second current_edge.p1)) { has_edge true; break; } } if (has_edge) { for (const auto v : tri.vertices) { if (v ! current_edge.p1 v ! current_edge.p2) { opposite_point v; break; } } break; } } // 在另一侧寻找候选点 std::vectorPoint candidates; for (const auto p : remaining_points) { double cross_current crossProduct(current_edge.p1, current_edge.p2, opposite_point); double cross_candidate crossProduct(current_edge.p1, current_edge.p2, p); if (cross_current * cross_candidate 0) { candidates.push_back(p); } } if (candidates.empty()) continue; // 选择最佳候选点 Point best_point findThirdPoint(current_edge.p1, current_edge.p2, candidates); if (best_point Point()) continue; // 创建新三角形 triangles.emplace_back(current_edge.p1, current_edge.p2, best_point); // 更新边状态 auto update_edge [](const Point a, const Point b) { Edge e{a, b}; if (e current_edge || current_edge e) { // 不是当前边 edge_usage[e]; edge_queue.push_back(e); } else { edge_usage[current_edge]; } }; update_edge(current_edge.p1, best_point); update_edge(current_edge.p2, best_point); } return triangles; } private: std::pairPoint, Point findClosestPair(std::vectorPoint points) { double min_dist std::numeric_limitsdouble::max(); std::pairPoint, Point closest_pair; for (size_t i 0; i points.size(); i) { for (size_t j i 1; j points.size(); j) { double dist points[i].distanceTo(points[j]); if (dist min_dist) { min_dist dist; closest_pair {points[i], points[j]}; } } } return closest_pair; } Point findThirdPoint(const Point p1, const Point p2, const std::vectorPoint candidates) { if (candidates.empty()) return Point(); double min_cos std::numeric_limitsdouble::max(); Point best_point; for (const auto p : candidates) { // 计算向量 double v1x p1.x - p.x, v1y p1.y - p.y; double v2x p2.x - p.x, v2y p2.y - p.y; // 计算点积和模 double dot v1x * v2x v1y * v2y; double mod std::sqrt(v1x*v1x v1y*v1y) * std::sqrt(v2x*v2x v2y*v2y); // 避免除零错误 if (mod 1e-10) continue; double cos_val dot / mod; if (cos_val min_cos) { min_cos cos_val; best_point p; } } return best_point; } double crossProduct(const Point a, const Point b, const Point p) { return (b.x - a.x) * (p.y - a.y) - (b.y - a.y) * (p.x - a.x); } };5. 性能优化与格式导出5.1 算法优化技巧空间索引加速使用网格或四叉树空间索引来加速最近邻搜索增量式更新维护一个活动边表避免重复计算并行计算对独立区域可并行处理# 使用网格空间索引的Python示例 class SpatialGrid: def __init__(self, points, cell_size): self.cell_size cell_size self.grid defaultdict(list) min_x min(p.x for p in points) min_y min(p.y for p in points) for p in points: cell_x int((p.x - min_x) / cell_size) cell_y int((p.y - min_y) / cell_size) self.grid[(cell_x, cell_y)].append(p) def query_neighbors(self, point, radius): min_x point.x - radius min_y point.y - radius max_x point.x radius max_y point.y radius candidates [] for cell_x in range(int(min_x/self.cell_size), int(max_x/self.cell_size)1): for cell_y in range(int(min_y/self.cell_size), int(max_y/self.cell_size)1): candidates.extend(self.grid.get((cell_x, cell_y), [])) return [p for p in candidates if min_x p.x max_x and min_y p.y max_y]5.2 导出为OBJ格式OBJ是一种简单的3D模型格式适合三角网格导出def export_to_obj(triangles, filename): with open(filename, w) as f: # 收集所有顶点 vertices [] vertex_index {} for tri in triangles: for v in tri.vertices: if (v.x, v.y, v.z) not in vertex_index: vertex_index[(v.x, v.y, v.z)] len(vertices) 1 vertices.append(v) # 写入顶点 for v in vertices: f.write(fv {v.x} {v.y} {v.z}\n) # 写入面 for tri in triangles: indices [vertex_index[(v.x, v.y, v.z)] for v in tri.vertices] f.write(ff {indices[0]} {indices[1]} {indices[2]}\n)C版本的OBJ导出void exportToObj(const std::vectorTriangle triangles, const std::string filename) { std::ofstream file(filename); if (!file.is_open()) return; std::mapstd::tupledouble, double, double, int vertex_index; std::vectorPoint vertices; // 收集顶点并建立索引 for (const auto tri : triangles) { for (const auto v : tri.vertices) { auto key std::make_tuple(v.x, v.y, v.z); if (vertex_index.find(key) vertex_index.end()) { vertex_index[key] vertices.size() 1; vertices.push_back(v); } } } // 写入顶点 for (const auto v : vertices) { file v v.x v.y v.z \n; } // 写入面 for (const auto tri : triangles) { file f ; for (const auto v : tri.vertices) { auto key std::make_tuple(v.x, v.y, v.z); file vertex_index[key] ; } file \n; } file.close(); }6. 两种语言实现对比特性Python实现C实现开发效率高代码简洁较低需要更多样板代码运行性能较慢适合小规模数据快适合大规模点云处理内存管理自动垃圾回收手动/智能指针管理多线程支持受GIL限制完整的多线程支持第三方库生态丰富NumPy, SciPy等较少但性能优化库多适合场景原型开发、小数据处理生产环境、高性能需求在实际项目中可以根据需求选择实现方式。对于需要快速验证算法或处理中小规模数据Python是更好的选择而对于性能要求高的生产环境C实现更为合适。
