fix:之前采样点集中在边缘，现在是表面均匀采样

brep2sdf/data/pre_process_by_mesh.py

@@ -25,7 +25,7 @@ from OCC.Core.StlAPI import StlAPI_Writer
 from OCC.Core.BRepAdaptor import BRepAdaptor_Surface
 from OCC.Core.gp import gp_Pnt, gp_Vec
 
-from brep2sdf.data.sampler import sample_sdf_points_and_normals, sample_face_points_brep, sample_edge_points_brep
+from brep2sdf.data.sampler import sample_sdf_points_and_normals, sample_face_points_brep, sample_edge_points_brep,sample_zero_surface_points_and_normals
 from brep2sdf.data.data import check_data_format
 from brep2sdf.data.utils import get_bbox, normalize, get_adjacency_info,batch_compute_normals
 from brep2sdf.utils.logger import logger
@@ -44,7 +44,8 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
     返回:
     dict: 包含以下键值对的字典:
         # 几何数据
-        'surf_wcs': np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(M, 3)的float32数组，表示面的点云坐标
+        'train_surf_ncs'  np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(M, 3)的float32数组，表示归一化后的面点云
+        'surf_wcs': np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(M, 3)的float32数组，表示面的点云坐标，很多是边缘点，不合适训练
         'edge_wcs': np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(num_edge_sample_points, 3)的float32数组，表示边的采样点坐标
         'surf_ncs': np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(M, 3)的float32数组，表示归一化后的面点云
         'edge_ncs': np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(num_edge_sample_points, 3)的float32数组，表示归一化后的边采样点
@@ -86,6 +87,7 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
     edge_explorer = TopExp_Explorer(shape, TopAbs_EDGE)
     vertex_explorer = TopExp_Explorer(shape, TopAbs_VERTEX)
     
+    #tarin_surf_pnts = []
     face_pnts = []
     edge_pnts = []
     corner_pnts = []
@@ -149,7 +151,7 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
         if face_idx < len(original_points):
             points = original_points[face_idx]
             target_points = points_per_face[face_idx]
-            
+            #tarin_surf_pnts.append(sample_face_points_brep(face, min_points=target_points))
             # 如果需要补充采样
             if len(points) < target_points:
                 try:
@@ -343,6 +345,7 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
     
     # 创建结果字典并确保所有数组都有正确的类型
     data = {
+        #'train_surf_ncs': np.array(train_surf_ncs, dtype=object),    # 保持对象数组
         'surf_wcs': np.array(surfs_wcs, dtype=object),    # 保持对象数组
         'edge_wcs': np.array(edges_wcs, dtype=object),    # 保持对象数组
         'surf_ncs': np.array(surfs_ncs, dtype=object),    # 保持对象数组
@@ -388,7 +391,10 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
 
             # 创建归一化 Trimesh
             vertices_wcs = trimesh_mesh.vertices.astype(np.float32)
-            vertices_ncs = (vertices_wcs - data['normalization_params']['center']) / data['normalization_params']['scale']
+            logger.debug(f"vertices_wcs:{vertices_wcs}")
+            logger.debug(f"center:{data['normalization_params']['center']},scale:{data['normalization_params']['scale']}")
+            vertices_ncs = (vertices_wcs - data['normalization_params']['center']) * data['normalization_params']['scale']
+            logger.debug(f"vertices_ncs:{vertices_ncs}")
             trimesh_mesh_ncs = trimesh.Trimesh(vertices=vertices_ncs, faces=trimesh_mesh.faces, process=False)
 
             if not trimesh_mesh_ncs.is_watertight:
@@ -397,11 +403,13 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
                 if not trimesh_mesh_ncs.is_watertight:
                     logger.warning(f"{step_path} 的归一化网格修复后仍不是 watertight。")
 
+            data["train_surf_ncs"] = sample_zero_surface_points_and_normals(trimesh_mesh_ncs, config.data.num_surf_points) # 归一化网格的顶点
+
         except Exception as e:
             logger.error(f"为 {step_path} 加载/处理 Trimesh 失败: {e}")
             trimesh_mesh = None
             trimesh_mesh_ncs = None
-
+    # 如果你需要归一化后的表面点
     # --- 计算表面点法线 ---
     if sample_normal_vector and trimesh_mesh_ncs is not None:
         logger.debug("计算表面点法线...")
@@ -435,6 +443,7 @@ def parse_solid(step_path,sample_normal_vector=False,sample_sdf_points=False):
 def process_single_step(step_path:str, output_path:str=None, sample_normal_vector=False, sample_sdf_points=False, timeout:int=300) -> dict:
     """处理单个STEP文件, 从 brep 2 pkl
     return data = {
+        'train_surf_ncs'  np.ndarray(dtype=object)     # 形状为(N,)的数组，每个元素是形状为(M, 3)的float32数组，表示归一化后的面点云
         'surf_wcs': np.array(surfs_wcs, dtype=object),    # 世界坐标系下的曲面几何数据(对象数组)
         'edge_wcs': np.array(edges_wcs, dtype=object),    # 世界坐标系下的边几何数据(对象数组)
         'surf_ncs': np.array(surfs_ncs, dtype=object),    # 归一化坐标系下的曲面几何数据(对象数组) 面归一化点云 [num_faces, num_surf_sample_points, 3]

brep2sdf/data/sampler.py

@@ -5,12 +5,13 @@ CAD模型处理脚本
 - 拓扑信息：面-边-顶点的邻接关系
 - 空间信息：包围盒数据
 """
-
+from typing import Optional
 import numpy as np
 from brep2sdf.utils.logger import logger
 import trimesh
 from trimesh.proximity import ProximityQuery
 
+
 # 导入OpenCASCADE相关库
 from OCC.Core.BRepAdaptor import BRepAdaptor_Surface, BRepAdaptor_Curve
 from OCC.Core.GeomLProp import GeomLProp_SLProps
@@ -185,10 +186,42 @@ def sample_edge_points_brep(edge, num_samples=50):
         return None
 
 
+def sample_zero_surface_points_and_normals(
+    trimesh_mesh_ncs: trimesh.Trimesh,
+    num_samples: int = 4096,
+) -> Optional[np.ndarray]:
+    """
+    从归一化 Trimesh 网格中均匀采样表面点及其对应法线，并附加 SDF=0，
+    返回形状为 (N, 7) 的数组：[x, y, z, nx, ny, nz, sdf=0.0]
+    
+    参数:
+        trimesh_mesh_ncs: 归一化的 Trimesh 网格
+        num_samples: 要采样的表面点数量
+    
+    返回:
+        points_with_normals_sdf: (N, 7) 的数组，包含坐标、法线、SDF=0
+    """
+    if not isinstance(trimesh_mesh_ncs, trimesh.Trimesh):
+        raise ValueError("输入必须是 trimesh.Trimesh 类型")
+
+    try:
+        # 表面采样点及对应的 face index
+        points, face_indices = trimesh_mesh_ncs.sample(num_samples, return_index=True)
+
+        # 获取每个点所在的三角面片的法线
+        normals = trimesh_mesh_ncs.face_normals[face_indices]
 
+        # 构造 sdf 标签为 0 的列
+        sdf_zeros = np.zeros((num_samples, 1), dtype=np.float32)
 
+        # 合并为 (N, 7) 的数组 [xyz, normal, sdf=0]
+        points_with_normals_sdf = np.hstack([points, normals, sdf_zeros], dtype=np.float32)
 
+        return points_with_normals_sdf
 
+    except Exception as e:
+        print(f"表面点采样失败: {e}")
+        return None
 
 def sample_sdf_points_and_normals(
     trimesh_mesh_ncs: trimesh.Trimesh,