alter

Former-commit-id: 0d9b33e7625efe7bee422d1514d8453ff689553d

test.py

@@ -0,0 +1,67 @@
+import cv2
+import numpy as np
+from matplotlib import pyplot as plt
+from scipy.signal import find_peaks
+
+cap = cv2.VideoCapture(1)  # 使用摄像头0，通常更稳定
+cap.set(cv2.CAP_PROP_FRAME_WIDTH, 640)  # 降低分辨率提高处理速度
+cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 480)
+
+# 预先创建图形窗口，避免重复创建
+fig, ax = plt.subplots(figsize=(10, 4))
+plt.ion()
+ax.set_title('Saturation Channel Histogram')
+ax.set_xlabel('Saturation Value')
+ax.set_ylabel('Pixel Count')
+ax.set_xlim(0, 255)
+
+while True:
+    ret, frame = cap.read()
+    if not ret:
+        print("Failed to grab frame")
+        break
+    
+    cv2.imshow("Camera Feed", frame)
+    
+    # 直接提取饱和度通道，避免完整HSV转换
+    hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
+    s = hsv[:, :, 1]
+    s = s[s > 0]  # 只保留非零饱和度值，减少噪声
+      # 使用更高效的直方图计算
+    hist = cv2.calcHist([s], [0], None, [256], [0, 256])
+    hist = hist.flatten()  # 转换为一维数组
+    
+    # 峰值检测 - 找到直方图中的峰值
+    peaks, properties = find_peaks(hist, 
+                                #  height=np.max(hist) * 0.1,  # 峰值高度至少是最大值的10%
+                                 distance=5,  # 峰值之间的最小距离
+                                 prominence=np.max(hist) * 0.05)  # 峰值的突出度
+    
+    # 清除旧数据并绘制新直方图
+    ax.clear()
+    ax.plot(hist, 'b-', linewidth=1)
+    
+    # 标注峰值
+    if len(peaks) > 0:
+        ax.text(0.5, 1.05, f'Found {len(peaks)} peaks')
+        ax.plot(peaks, hist[peaks], 'ro', markersize=8, label=f'Peaks ({len(peaks)})')
+        # 在峰值处添加文字标注
+        for i, peak in enumerate(peaks):
+            ax.annotate(f'Peak {i+1}\n({peak}, {int(hist[peak])})', 
+                       xy=(peak, hist[peak]), 
+                       xytext=(peak, hist[peak] + np.max(hist) * 0.1),
+                       ha='center', va='bottom',
+                       bbox=dict(boxstyle='round,pad=0.3', facecolor='yellow', alpha=0.7),
+                       arrowprops=dict(arrowstyle='->', color='red'))
+
+    
+    plt.draw()
+    plt.pause(0.1)  # 确保图形更新
+    
+    key = cv2.waitKey(1) & 0xFF
+    if key == ord('q'):
+        break
+
+cap.release()
+cv2.destroyAllWindows()
+plt.close('all')

tmp.py

@@ -0,0 +1,80 @@
+import numpy as np
+
+class DistributionChangeDetector:
+    def __init__(self, baseline_windows: list[np.ndarray]):
+        """
+        参数 baseline_windows: List of arrays，代表初始稳定期的多个窗口
+        """
+        self.baseline = self._compute_baseline(baseline_windows)
+
+    def _compute_stats(self, window: np.ndarray) -> tuple[float, float, float]:
+        """返回 (P_under30, std, mode)"""
+        p_under30 = np.mean(window < 30)
+        std = np.std(window, ddof=1)
+        
+        # 快速估计众数：最大 bin 的中心
+        hist, bin_edges = np.histogram(window, bins=50)
+        max_bin_index = np.argmax(hist)
+        mode_est = (bin_edges[max_bin_index] + bin_edges[max_bin_index + 1]) / 2
+        return p_under30, std, mode_est
+
+    def _compute_baseline(self, windows: list[np.ndarray]) -> tuple[np.ndarray, np.ndarray]:
+        """
+        返回 baseline 向量 (P0, σ0, mode0) 和对应标准差（用于归一化）
+        """
+        stats = np.array([self._compute_stats(w) for w in windows])
+        mean = stats.mean(axis=0)
+        std = stats.std(axis=0) + 1e-6  # 防止除0
+        return mean, std
+
+    def update(self, window: np.ndarray) -> float:
+        """
+        输入：当前窗口数据（长度 = 窗口大小）
+        输出：变化分数（越大表示分布越偏离基准）
+        """
+        x = np.array(self._compute_stats(window))
+        mean, std = self.baseline
+        norm_diff = (x - mean) / std
+        change_score = np.linalg.norm(norm_diff)
+        return float(change_score)
+
+
+import cv2
+def gen_data():
+    cap = cv2.VideoCapture()
+    cap.open(1)
+    while True:
+        ret, frame = cap.read()
+        cv2.imshow("Camera Feed", frame)
+        if not ret:
+            break
+        hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
+        s = hsv[:, :, 1]  # 直接提取饱和度通道
+        s = s[s > 0]  # 只保留非零饱和度值，减少噪声
+        yield s
+        if cv2.waitKey(1) & 0xFF == ord('a'):
+            break
+
+
+gen = gen_data()
+baseline_data = [gen.__next__() for _ in range(5)]  # 获取10个窗口作为基线
+
+det = DistributionChangeDetector(baseline_data)
+
+
+
+results = []
+for x in gen:
+    out = det.update(x)
+    if out is not None:
+        results.append(out)
+
+
+# 作图查看
+import matplotlib.pyplot as plt
+plt.plot(results, label="ChangeScore")
+plt.xlabel("Window index")
+plt.ylabel("Score")
+plt.title("Streaming Change Detection")
+plt.legend()
+plt.show()

tmp2.py

@@ -0,0 +1,144 @@
+import numpy as np
+import cv2
+import matplotlib.pyplot as plt
+from matplotlib.animation import FuncAnimation
+from collections import deque
+import threading
+import time
+
+class DistributionChangeDetector:
+    def __init__(self, baseline_windows: list[np.ndarray]):
+        """
+        参数 baseline_windows: List of arrays，代表初始稳定期的多个窗口
+        """
+        self.baseline = self._compute_baseline(baseline_windows)
+
+    def _compute_stats(self, window: np.ndarray) -> tuple[float, float, float]:
+        """返回 (P_under30, std, mode)"""
+        p_under30 = np.mean(window < 30)
+        std = np.std(window, ddof=1)
+        
+        # 快速估计众数：最大 bin 的中心
+        hist, bin_edges = np.histogram(window, bins=50)
+        max_bin_index = np.argmax(hist)
+        mode_est = (bin_edges[max_bin_index] + bin_edges[max_bin_index + 1]) / 2
+        return p_under30, std, mode_est
+
+    def _compute_baseline(self, windows: list[np.ndarray]) -> tuple[np.ndarray, np.ndarray]:
+        """
+        返回 baseline 向量 (P0, σ0, mode0) 和对应标准差（用于归一化）
+        """
+        stats = np.array([self._compute_stats(w) for w in windows])
+        mean = stats.mean(axis=0)
+        std = stats.std(axis=0) + 1e-6  # 防止除0
+        return mean, std
+
+    def update(self, window: np.ndarray) -> float:
+        """
+        输入：当前窗口数据（长度 = 窗口大小）
+        输出：变化分数（越大表示分布越偏离基准）
+        """
+        x = np.array(self._compute_stats(window))
+        mean, std = self.baseline
+        norm_diff = (x - mean) / std
+        change_score = np.linalg.norm(norm_diff)
+        return float(change_score)
+
+def hsv_score(s:np.ndarray):
+    mask = s>30
+    tot = len(mask)
+    val = np.sum(mask)
+    rate = val/tot
+    return rate
+
+class RealTimePlotter:
+    def __init__(self, max_points=200):
+        self.max_points = max_points
+        self.scores = deque(maxlen=max_points)
+        self.scores2 = deque(maxlen=max_points)
+        self.times = deque(maxlen=max_points)
+        self.start_time = time.time()
+        
+        # 设置图形
+        plt.ion()  # 打开交互模式
+        self.fig, (self.ax,self.ax2) = plt.subplots(1,2,figsize=(10, 6))
+        self.line, = self.ax.plot([], [], 'b-', linewidth=2)
+        self.line2, = self.ax2.plot([], [], 'b-', linewidth=2)
+        self.ax.set_xlabel('Time (s)')
+        self.ax.set_ylabel('Change Score')
+        self.ax.set_title('Real-time Distribution Change Detection')
+        self.ax.grid(True)
+        self.ax2.grid(True)
+        
+    def update_plot(self, score,s_score):
+        current_time = time.time() - self.start_time
+        self.scores.append(score)
+        self.scores2.append(s_score)
+        self.times.append(current_time)
+        
+        # 更新数据
+        self.line.set_data(list(self.times), list(self.scores))
+        self.line2.set_data(list(self.times), list(self.scores2))
+        
+        # 自动调整坐标轴
+        if len(self.times) > 1:
+            self.ax.set_xlim(min(self.times), max(self.times))
+            self.ax2.set_xlim(min(self.times), max(self.times))
+            self.ax.set_ylim(0,100)
+            # self.ax.set_ylim(min(self.scores) * 0.95, max(self.scores) * 1.05)
+            self.ax2.set_ylim(0,1)
+        
+        # 刷新图形
+        self.fig.canvas.draw()
+        self.fig.canvas.flush_events()
+
+
+def gen_data():
+    cap = cv2.VideoCapture(1)
+    while True:
+        ret, frame = cap.read()
+        if not ret:
+            break
+        cv2.imshow("Camera Feed", frame)
+        hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
+        s = hsv[:, :, 1]  # 直接提取饱和度通道
+        s = s[s > 0]  # 只保留非零饱和度值，减少噪声
+        yield s
+        if cv2.waitKey(1) & 0xFF == ord('a'):
+            break
+    cap.release()
+    cv2.destroyAllWindows()
+
+
+def main():
+    # 初始化数据生成器
+    gen = gen_data()
+    
+    # 获取基线数据
+    print("收集基线数据...")
+    baseline_data = [next(gen) for _ in range(30*5)]
+    
+    # 初始化检测器和绘图器
+    det = DistributionChangeDetector(baseline_data)
+    plotter = RealTimePlotter()
+    
+    print("开始实时检测和绘图...")
+    
+    try:
+        for x in gen:
+            score = det.update(x)
+            score2 = hsv_score(x)
+            plotter.update_plot(score,score2)
+            
+            # 小延时以控制更新频率
+            time.sleep(0.01)
+            
+    except KeyboardInterrupt:
+        print("停止检测")
+    finally:
+        plt.ioff()  # 关闭交互模式
+        plt.show()  # 保持最终图形显示
+
+
+if __name__ == "__main__":
+    main()