上图先
python
# -*- coding: utf-8 -*-
import cv2
import mediapipe as mp
import numpy as np
import time
import sys
import os
import tempfile
import subprocess
# 解决中文显示问题 - 使用Pillow确保中文正确显示
def cv2_puttext_chinese(img, text, position, font_scale, color, thickness):
"""
使用Pillow库在OpenCV图像上显示中文
"""
try:
from PIL import Image, ImageDraw, ImageFont
# 确保颜色格式正确
if isinstance(color, tuple) and len(color) == 3:
# OpenCV是BGR格式，需要转换为RGB
color_rgb = (color[2], color[1], color[0])
else:
color_rgb = (255, 255, 255)
# 将OpenCV图像转换为PIL图像
img_pil = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_pil)
# 确定字体大小
font_size = int(font_scale * 20) # 调整比例以匹配cv2.putText
# 尝试使用系统中常见的中文字体
font_paths = [
"C:/Windows/Fonts/simhei.ttf", # 黑体
"C:/Windows/Fonts/msyh.ttc", # 微软雅黑
"C:/Windows/Fonts/msyhbd.ttc", # 微软雅黑粗体
"C:/Windows/Fonts/simsun.ttc", # 宋体
"C:/Windows/Fonts/arial.ttf", # 英文备选
]
font = None
for font_path in font_paths:
try:
if os.path.exists(font_path):
font = ImageFont.truetype(font_path, font_size)
break
except Exception:
continue
# 如果没有找到合适的字体，使用默认字体
if font is None:
font = ImageFont.load_default()
# 绘制中文文本
draw.text(position, text, font=font, fill=color_rgb)
# 转换回OpenCV格式
img = cv2.cvtColor(np.array(img_pil), cv2.COLOR_RGB2BGR)
return img
except ImportError:
# 如果PIL不可用，使用简单的英文替代
english_map = {
"右手抹鼻子": "Right hand to nose",
"左手摸头发": "Left hand to hair",
"双手击掌": "Hands clap",
"SOP完成！": "SOP Complete!",
"剩余时间": "Time left",
"秒": "s"
}
for chinese, english in english_map.items():
text = text.replace(chinese, english)
cv2.putText(img, text, position, cv2.FONT_HERSHEY_SIMPLEX,
font_scale, color, thickness, cv2.LINE_AA)
return img
except Exception as e:
# 其他错误情况下，使用英文替代
english_map = {
"右手抹鼻子": "Right hand to nose",
"左手摸头发": "Left hand to hair",
"双手击掌": "Hands clap",
"SOP完成！": "SOP Complete!",
"剩余时间": "Time left",
"秒": "s"
}
for chinese, english in english_map.items():
text = text.replace(chinese, english)
cv2.putText(img, text, position, cv2.FONT_HERSHEY_SIMPLEX,
font_scale, color, thickness, cv2.LINE_AA)
return img
class SOPDetector:
def __init__(self):
# 初始化MediaPipe姿势检测
self.mp_pose = mp.solutions.pose
self.pose = self.mp_pose.Pose(
static_image_mode=False,
model_complexity=1,
smooth_landmarks=True,
enable_segmentation=False,
smooth_segmentation=True,
min_detection_confidence=0.5,
min_tracking_confidence=0.5
)
self.mp_drawing = mp.solutions.drawing_utils
# 定义关键点索引
self.NOSE = 0
self.LEFT_EAR = 7
self.RIGHT_EAR = 8
self.LEFT_WRIST = 15
self.RIGHT_WRIST = 16
# SOP状态
self.current_step = 0
self.steps_completed = [False, False, False]
self.step_start_time = time.time()
self.step_timeout = 10 # 每个步骤的超时时间（秒）
# 动作持续时间参数
self.action_duration_required = 1.5 # 每个动作需要维持的时间（秒）
self.action_start_time = None # 记录当前动作开始的时间
# 动作检测阈值
self.NOSE_TOUCH_DISTANCE = 0.1
self.HAIR_TOUCH_DISTANCE = 0.15
self.HANDS_CLAP_DISTANCE = 0.1
# 显示控制
self.show_skeleton = True # 是否显示骨架
self.black_background = False # 是否显示黑色背景
self.window_name = 'SOP Detector' # 窗口名称
# 操作日志
self.logs = [] # 存储日志记录
self.max_log_lines = 8 # 最大显示日志行数
def calculate_distance(self, point1, point2):
"""计算两个关键点之间的欧几里得距离"""
return np.sqrt((point1.x - point2.x)**2 + (point1.y - point2.y)**2)
def add_log(self, event):
"""添加操作日志"""
timestamp = time.strftime("%H:%M:%S", time.localtime())
log_entry = f"[{timestamp}] {event}"
self.logs.append(log_entry)
# 保持日志行数不超过最大限制
if len(self.logs) > self.max_log_lines:
self.logs = self.logs[-self.max_log_lines:]
# 同时在控制台输出
print(log_entry)
def detect_step1_right_hand_nose(self, landmarks):
"""检测步骤1：右手抹鼻子"""
nose = landmarks[self.NOSE]
right_wrist = landmarks[self.RIGHT_WRIST]
distance = self.calculate_distance(nose, right_wrist)
return distance < self.NOSE_TOUCH_DISTANCE
def detect_step2_left_hand_hair(self, landmarks):
"""检测步骤2：左手摸头发"""
left_ear = landmarks[self.LEFT_EAR]
left_wrist = landmarks[self.LEFT_WRIST]
distance = self.calculate_distance(left_ear, left_wrist)
return distance < self.HAIR_TOUCH_DISTANCE
def detect_step3_hands_clap(self, landmarks):
"""检测步骤3：左手和右手击掌"""
left_wrist = landmarks[self.LEFT_WRIST]
right_wrist = landmarks[self.RIGHT_WRIST]
distance = self.calculate_distance(left_wrist, right_wrist)
return distance < self.HANDS_CLAP_DISTANCE
def check_timeout(self):
"""检查当前步骤是否超时"""
if time.time() - self.step_start_time > self.step_timeout:
print(f"步骤 {self.current_step + 1} 超时，请重新开始")
self.current_step = 0
self.steps_completed = [False, False, False]
self.step_start_time = time.time()
return True
return False
def update_sop_state(self, landmarks):
"""更新SOP状态"""
if self.check_timeout():
return
# 检查当前步骤是否完成
if self.current_step == 0 and not self.steps_completed[0]:
if self.detect_step1_right_hand_nose(landmarks):
if self.action_start_time is None:
# 动作开始，记录时间
self.action_start_time = time.time()
self.add_log("检测到右手抹鼻子动作...")
elif time.time() - self.action_start_time >= self.action_duration_required:
# 动作维持了足够时间，标记完成
self.steps_completed[0] = True
self.current_step = 1
self.step_start_time = time.time()
self.action_start_time = None # 重置动作计时器
self.add_log("步骤1完成：右手抹鼻子")
else:
# 动作中断，重置计时器
if self.action_start_time is not None:
self.action_start_time = None
self.add_log("右手抹鼻子动作中断，请重新执行")
elif self.current_step == 1 and not self.steps_completed[1]:
if self.detect_step2_left_hand_hair(landmarks):
if self.action_start_time is None:
# 动作开始，记录时间
self.action_start_time = time.time()
self.add_log("检测到左手摸头发动作...")
elif time.time() - self.action_start_time >= self.action_duration_required:
# 动作维持了足够时间，标记完成
self.steps_completed[1] = True
self.current_step = 2
self.step_start_time = time.time()
self.action_start_time = None # 重置动作计时器
self.add_log("步骤2完成：左手摸头发")
else:
# 动作中断，重置计时器
if self.action_start_time is not None:
self.action_start_time = None
self.add_log("左手摸头发动作中断，请重新执行")
elif self.current_step == 2 and not self.steps_completed[2]:
if self.detect_step3_hands_clap(landmarks):
if self.action_start_time is None:
# 动作开始，记录时间
self.action_start_time = time.time()
self.add_log("检测到双手击掌动作...")
elif time.time() - self.action_start_time >= self.action_duration_required:
# 动作维持了足够时间，标记完成
self.steps_completed[2] = True
self.current_step = 3
self.action_start_time = None # 重置动作计时器
self.add_log("步骤3完成：左手和右手击掌")
self.add_log("恭喜！所有SOP步骤完成！")
else:
# 动作中断，重置计时器
if self.action_start_time is not None:
self.action_start_time = None
self.add_log("双手击掌动作中断，请重新执行")
def draw_landmarks_and_info(self, image, landmarks):
"""绘制关键点和状态信息"""
# 创建黑色背景图像
height, width, _ = image.shape
black_bg = np.zeros((height, width, 3), dtype=np.uint8)
# 选择显示背景
display_image = black_bg if self.black_background else image.copy()
# 绘制姿势关键点（如果启用）
if self.show_skeleton:
self.mp_drawing.draw_landmarks(
display_image, landmarks, self.mp_pose.POSE_CONNECTIONS)
# 添加状态信息
# 当前步骤信息
step_texts = [
"S1: 右手抹鼻子",
"S2: 左手摸头发",
"S3: 双手击掌"
]
for i, text in enumerate(step_texts):
color = (0, 255, 0) if self.steps_completed[i] else (0, 0, 255)
if i == self.current_step and not self.steps_completed[i]:
color = (255, 255, 0) # 黄色表示当前步骤
# 使用中文显示函数
display_image = cv2_puttext_chinese(display_image, text, (10, 30 + i*30),
0.7, color, 2)
# 显示动作持续时间进度
if self.current_step < 3 and not self.steps_completed[self.current_step] and self.action_start_time is not None:
elapsed_action_time = time.time() - self.action_start_time
progress = min(elapsed_action_time / self.action_duration_required, 1.0)
# 绘制进度条
bar_width = 200
bar_height = 20
x_start = 10
y_start = 120
# 背景
cv2.rectangle(display_image, (x_start, y_start), (x_start + bar_width, y_start + bar_height), (50, 50, 50), -1)
# 进度
cv2.rectangle(display_image, (x_start, y_start), (x_start + int(bar_width * progress), y_start + bar_height), (0, 255, 255), -1)
# 边框
cv2.rectangle(display_image, (x_start, y_start), (x_start + bar_width, y_start + bar_height), (255, 255, 255), 2)
# 显示进度文字
progress_text = f"动作持续中: {elapsed_action_time:.1f}/{self.action_duration_required:.1f}秒"
display_image = cv2_puttext_chinese(display_image, progress_text, (10, 155), 0.6, (255, 255, 255), 2)
# 剩余时间
if self.current_step < 3:
remaining_time = self.step_timeout - (time.time() - self.step_start_time)
time_text = f"剩余时间: {max(0, int(remaining_time))}秒"
display_image = cv2_puttext_chinese(display_image, time_text, (width - 200, 30),
0.7, (255, 255, 255), 2)
# 完成状态
if all(self.steps_completed):
display_image = cv2_puttext_chinese(display_image, "SOP完成！", (width//2 - 100, 50),
1.5, (0, 255, 0), 3)
# 添加操作日志显示
log_height = 180 # 增加日志区域高度
log_start_y = display_image.shape[0] - log_height
log_text_color = (0, 0, 0) # 黑色文字
# 绘制日志背景
log_bg_top_left = (10, log_start_y - 10)
log_bg_bottom_right = (display_image.shape[1] - 10, display_image.shape[0] - 10)
cv2.rectangle(display_image, log_bg_top_left, log_bg_bottom_right, (255, 255, 255), -1) # 白色背景
cv2.rectangle(display_image, log_bg_top_left, log_bg_bottom_right, (0, 0, 0), 2) # 黑色边框
# 绘制日志标题
display_image = cv2_puttext_chinese(display_image, "操作日志", (15, log_start_y), 1.2, (0, 0, 255), 2) # 红色标题
# 绘制日志内容
for i, log in enumerate(self.logs):
y_pos = log_start_y + 30 + (i * 20)
display_image = cv2_puttext_chinese(display_image, log, (15, y_pos), 0.8, log_text_color, 1)
# 添加控制提示（放在日志区域内）
control_text = "控制提示: 空格键-骨架显示, B键-黑色背景, Q键-退出"
display_image = cv2_puttext_chinese(display_image, control_text, (15, display_image.shape[0] - 20), 0.5, (0, 128, 0), 1)
return display_image
def run_detection(self):
"""运行SOP检测"""
cap = cv2.VideoCapture(0)
self.add_log("SOP实时侦测系统启动")
self.add_log("请按以下顺序完成动作：")
self.add_log("1. 右手抹鼻子")
self.add_log("2. 左手摸头发")
self.add_log("3. 双手击掌")
self.add_log("按'q'键退出程序")
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# 转换BGR到RGB
rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# 姿势检测
results = self.pose.process(rgb_frame)
display_frame = frame.copy()
if results.pose_landmarks:
# 更新SOP状态
self.update_sop_state(results.pose_landmarks.landmark)
# 绘制关键点和信息
display_frame = self.draw_landmarks_and_info(frame, results.pose_landmarks)
# 显示帧
cv2.imshow(self.window_name, display_frame)
# 键盘事件处理
key = cv2.waitKey(1) & 0xFF
# 按'q'退出
if key == ord('q'):
break
# 按空格键切换骨架显示
elif key == ord(' '):
self.show_skeleton = not self.show_skeleton
status = "开启" if self.show_skeleton else "关闭"
self.add_log(f"骨架显示已{status}")
# 按'b'键切换黑色背景
elif key == ord('b') or key == ord('B'):
self.black_background = not self.black_background
status = "开启" if self.black_background else "关闭"
self.add_log(f"黑色背景已{status}")
cap.release()
cv2.destroyAllWindows()
def main():
detector = SOPDetector()
detector.run_detection()
if __name__ == "__main__":
main()