1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
| from dataclasses import dataclass
from typing import List, Tuple, Optional
import numpy as np
@dataclass
class RadarPoint:
"""雷达点云数据"""
x: float
y: float
z: float
velocity: float
snr: float
timestamp: float
@dataclass
class ToFDepthFrame:
"""ToF 深度帧"""
depth_map: np.ndarray
amplitude: np.ndarray
timestamp: float
@dataclass
class RGBFrame:
"""RGB/IR 图像帧"""
image: np.ndarray
is_infrared: bool
timestamp: float
@dataclass
class FusedOccupantData:
"""融合后的乘员数据"""
position_3d: Tuple[float, float, float]
velocity: Optional[float]
depth_map_roi: np.ndarray
visual_features: Optional[np.ndarray]
detection_confidence: float
timestamp: float
class SensorFusionPipeline:
"""
传感器融合管线
整合雷达、ToF、视觉数据
"""
def __init__(self, config: dict):
self.config = config
self.radar_to_camera_extrinsics = config.get('radar_to_camera_extrinsics', np.eye(4))
self.tof_to_camera_extrinsics = config.get('tof_to_camera_extrinsics', np.eye(4))
self.radar_buffer = []
self.tof_buffer = []
self.rgb_buffer = []
self.sync_threshold_ms = 33
def process_frame(
self,
radar_data: List[RadarPoint],
tof_frame: ToFDepthFrame,
rgb_frame: RGBFrame
) -> List[FusedOccupantData]:
"""
处理一帧数据
Args:
radar_data: 雷达点云
tof_frame: ToF 深度帧
rgb_frame: RGB 图像
Returns:
occupants: 检测到的乘员列表
"""
if not self._check_temporal_sync(radar_data, tof_frame, rgb_frame):
return []
radar_clusters = self._cluster_radar_points(radar_data)
tof_occupants = self._detect_from_tof(tof_frame)
visual_detections = self._detect_from_rgb(rgb_frame)
fused_occupants = self._fuse_detections(
radar_clusters, tof_occupants, visual_detections
)
return fused_occupants
def _check_temporal_sync(
self,
radar_data: List[RadarPoint],
tof_frame: ToFDepthFrame,
rgb_frame: RGBFrame
) -> bool:
"""检查时间同步"""
if not radar_data or tof_frame is None or rgb_frame is None:
return False
radar_time = radar_data[0].timestamp
tof_time = tof_frame.timestamp
rgb_time = rgb_frame.timestamp
max_diff = max(abs(radar_time - tof_time),
abs(radar_time - rgb_time),
abs(tof_time - rgb_time))
return max_diff < self.sync_threshold_ms / 1000.0
def _cluster_radar_points(
self,
points: List[RadarPoint]
) -> List[dict]:
"""雷达点云聚类"""
if not points:
return []
from sklearn.cluster import DBSCAN
positions = np.array([[p.x, p.y, p.z] for p in points])
clustering = DBSCAN(eps=0.3, min_samples=5).fit(positions)
labels = clustering.labels_
clusters = []
for label in set(labels):
if label == -1:
continue
cluster_points = [p for i, p in enumerate(points) if labels[i] == label]
center = np.mean([[p.x, p.y, p.z] for p in cluster_points], axis=0)
avg_velocity = np.mean([p.velocity for p in cluster_points])
is_breathing = self._detect_breathing(cluster_points)
clusters.append({
'center': center,
'point_count': len(cluster_points),
'avg_velocity': avg_velocity,
'is_breathing': is_breathing,
'points': cluster_points
})
return clusters
def _detect_breathing(self, points: List[RadarPoint]) -> bool:
"""检测呼吸运动"""
velocities = [abs(p.velocity) for p in points]
avg_velocity = np.mean(velocities)
return 0.005 < avg_velocity < 0.1
def _detect_from_tof(self, frame: ToFDepthFrame) -> List[dict]:
"""ToF 深度检测"""
depth_map = frame.depth_map
valid_mask = (depth_map > 0.3) & (depth_map < 3.0)
from scipy.ndimage import label
labeled, num_features = label(valid_mask)
detections = []
for i in range(1, num_features + 1):
mask = labeled == i
ys, xs = np.where(mask)
depths = depth_map[mask]
center_x = np.mean(xs)
center_y = np.mean(ys)
center_z = np.mean(depths)
bbox = [min(xs), min(ys), max(xs), max(ys)]
detections.append({
'position': (center_x, center_y, center_z),
'bbox': bbox,
'pixel_count': len(xs),
'depth_map_roi': depth_map[bbox[1]:bbox[3], bbox[0]:bbox[2]]
})
return detections
def _detect_from_rgb(self, frame: RGBFrame) -> List[dict]:
"""视觉检测"""
return []
def _fuse_detections(
self,
radar_clusters: List[dict],
tof_detections: List[dict],
visual_detections: List[dict]
) -> List[FusedOccupantData]:
"""融合检测结果"""
fused = []
for tof_det in tof_detections:
tof_pos = tof_det['position']
matched_radar = None
min_dist = float('inf')
for radar_cluster in radar_clusters:
radar_pos = radar_cluster['center']
dist = np.linalg.norm(np.array(tof_pos) - np.array(radar_pos))
if dist < min_dist and dist < 0.5:
min_dist = dist
matched_radar = radar_cluster
occupant = FusedOccupantData(
position_3d=tof_pos,
velocity=matched_radar['avg_velocity'] if matched_radar else None,
depth_map_roi=tof_det['depth_map_roi'],
visual_features=None,
detection_confidence=0.9 if matched_radar else 0.7,
timestamp=0.0
)
fused.append(occupant)
return fused
|