h5ad generation

embryo-backend/Data/glbTrans.py

@@ -3,7 +3,6 @@ import trimesh
 import pandas as pd
 import numpy as np
 import anndata as ad
-from collections import defaultdict
 
 # ==== 参数 ====
 script_dir = os.path.dirname(os.path.realpath(__file__))
@@ -30,32 +29,26 @@ true_markers = {
 
 # ==== 采样函数 ====
 def sample_mesh(mesh, label, n_samples):
-    """采样点云 + 颜色 + 标签"""
-    if use_surface_sampling:
-        points, face_idx = trimesh.sample.sample_surface(mesh, n_samples)
-        if hasattr(mesh.visual, "vertex_colors") and len(mesh.visual.vertex_colors) == len(mesh.vertices):
-            face_colors = mesh.visual.vertex_colors[mesh.faces]
-            colors = face_colors.mean(axis=1)[face_idx, :3]
-        else:
-            mat_color = np.array([200, 200, 200])
-            colors = np.tile(mat_color, (n_samples, 1))
+    if len(mesh.vertices) == 0:
+        print(f"⚠️ {label} 无顶点，跳过")
+        return np.zeros((0,3)), np.zeros((0,3)), np.array([])
+
+    if mesh.faces is None or len(mesh.faces) == 0 or not use_surface_sampling:
+        idx = np.random.choice(len(mesh.vertices), min(n_samples, len(mesh.vertices)), replace=False)
+        points = mesh.vertices[idx]
     else:
-        verts = mesh.vertices
-        if len(verts) > n_samples:
-            idx = np.random.choice(len(verts), n_samples, replace=False)
-            verts = verts[idx]
-        else:
-            idx = np.arange(len(verts))
-        if hasattr(mesh.visual, "vertex_colors") and len(mesh.visual.vertex_colors) == len(mesh.vertices):
-            colors = mesh.visual.vertex_colors[idx, :3]
-        else:
-            mat_color = np.array([200, 200, 200])
-            colors = np.tile(mat_color, (len(verts), 1))
-        points = verts
+        points, face_idx = trimesh.sample.sample_surface(mesh, n_samples)
+
+    if hasattr(mesh.visual, "vertex_colors") and len(mesh.visual.vertex_colors) == len(mesh.vertices):
+        colors = mesh.visual.vertex_colors[:len(points), :3]
+    else:
+        mat_color = np.array([200,200,200])
+        colors = np.tile(mat_color, (len(points),1))
+
     labels = np.array([label] * len(points))
     return points, colors, labels
 
-# ==== Step 1: 计算全局最大边长 ====
+# ==== Step 1: 计算全局边界 ====
 all_bounds = []
 for glb_name in glb_files:
     glb_path = os.path.join(script_dir, glb_name)
@@ -74,7 +67,7 @@ global_max = np.max(all_bounds[:,1,:], axis=0)
 global_size = np.max(global_max - global_min)
 print(f"🌍 统一缩放基准: global_size={global_size}")
 
-# ==== Step 2: 处理单个 GLB（应用全局变换） ====
+# ==== Step 2: 处理单个 GLB ====
 def process_glb(glb_path, sample_name):
     scene = trimesh.load(glb_path)
     all_points, all_colors, all_labels = [], [], []
@@ -82,23 +75,32 @@ def process_glb(glb_path, sample_name):
     if isinstance(scene, trimesh.Scene):
         total_points = sum(len(g.vertices) for g in scene.geometry.values())
         for name, geom in scene.geometry.items():
-            ratio = len(geom.vertices) / total_points
-            n_samples = max(1, int(N_total * ratio))
+            if len(geom.vertices) == 0:
+                print(f"⚠️ {sample_name}: 子网格 {name} 无顶点，跳过")
+                continue
 
-            # ✅ 保护性获取变换矩阵
+            ratio = len(geom.vertices) / total_points if total_points > 0 else 0
+            n_samples = max(100, int(N_total * ratio))  # ✅ 最少100点
+
+            # ✅ 自动检测 & 兼容修复
             try:
                 transform = scene.graph.get(name)[0]
+                verts_world = trimesh.transform_points(geom.vertices, transform)
             except Exception:
-                print(f"⚠️ {sample_name}: 子网格 {name} 没有变换路径，使用单位矩阵")
-                transform = np.eye(4)
+                # 判断顶点是否已经全局化
+                local_bounds = geom.bounds
+                local_size = np.max(local_bounds[1] - local_bounds[0])
+                if local_size > 0.5 * global_size:  # 说明已在全局坐标
+                    print(f"ℹ️ {sample_name}: 子网格 {name} 似乎已是全局坐标 → 直接使用")
+                    verts_world = geom.vertices.copy()
+                else:
+                    print(f"⚠️ {sample_name}: 子网格 {name} 无变换路径且看似局部坐标 → 仍使用原始顶点（可能错位）")
+                    verts_world = geom.vertices.copy()
 
-            # ✅ 应用全局变换
-            verts_world = trimesh.transform_points(geom.vertices, transform)
-
-            # ✅ 生成全局 mesh 进行采样
             mesh_world = trimesh.Trimesh(vertices=verts_world, faces=geom.faces, process=False)
             p, c, l = sample_mesh(mesh_world, name, n_samples)
-
+            if len(p) == 0:
+                continue
             all_points.append(p)
             all_colors.append(c)
             all_labels.append(l)
@@ -108,15 +110,14 @@ def process_glb(glb_path, sample_name):
         all_colors.append(c)
         all_labels.append(l)
 
+    if len(all_points) == 0:
+        raise RuntimeError(f"❌ {sample_name}: 没有有效点被采样！")
+
     points = np.vstack(all_points)
     colors = np.vstack(all_colors)
     labels = np.concatenate(all_labels)
 
-    if len(points) > N_total:
-        idx = np.random.choice(len(points), N_total, replace=False)
-        points, colors, labels = points[idx], colors[idx], labels[idx]
-
-    # ✅ 一次性中心化 + 统一缩放
+    # ✅ 中心化 + 统一缩放
     center = points.mean(axis=0)
     points_centered = points - center
     points_scaled = points_centered / global_size
@@ -127,10 +128,9 @@ def process_glb(glb_path, sample_name):
 
     csv_file = os.path.join(script_dir, f"{sample_name}_point_cloud_30000_centered_scaled.csv")
     df.to_csv(csv_file, index=False)
-    print(f"✅ 已导出 {csv_file} (修正组织错位, 中心化 & 统一比例)")
+    print(f"✅ 已导出 {csv_file} (自动检测 + 兼容修复)")
     return df
 
-
 # ==== Step 3: 生成 h5ad ====
 def create_h5ad(df, sample_name):
     points = df[['x','y','z']].to_numpy()