python摄像头一直打开，知道其轮廓周长大于1/2时，开始采集从这一时刻开始到3秒后结束或者前100帧的所有图像左极值点，并将所有左极值点垂直像素进行排序，求出最大值和最小值之差

这里给出一个可能的实现：

import cv2
import time

# 打开摄像头
cap = cv2.VideoCapture(0)

# 初始化相关变量
start_time = None
left_points = []
max_y = -1
min_y = 100000

while True:
    # 读取一帧图像
    ret, frame = cap.read()
    
    # 转为灰度图并进行阈值化处理
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    _, thresh = cv2.threshold(gray, 100, 255, cv2.THRESH_BINARY)
    
    # 计算轮廓并找到最长的轮廓
    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    max_contour = None
    max_contour_length = -1
    for contour in contours:
        length = cv2.arcLength(contour, True)
        if length > max_contour_length:
            max_contour = contour
            max_contour_length = length
    
    # 如果轮廓周长大于1/2，则开始采集左极值点
    if max_contour_length > gray.shape[1] // 2:
        if start_time is None:
            start_time = time.time()
        if len(left_points) < 100 and time.time() - start_time < 3:
            leftmost = tuple(max_contour[max_contour[:,:,0].argmin()][0])
            left_points.append(leftmost[1])
            if leftmost[1] > max_y:
                max_y = leftmost[1]
            if leftmost[1] < min_y:
                min_y = leftmost[1]
        else:
            break
            
    # 显示当前帧
    cv2.imshow('frame', frame)
    
    # 按下q键退出
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# 释放摄像头并关闭所有窗口
cap.release()
cv2.destroyAllWindows()

# 对左极值点垂直像素进行排序并求出最大值和最小值之差
left_points.sort()
diff = left_points[-1] - left_points[0]
print('max_y:', max_y)
print('min_y:', min_y)
print('diff:', diff)

注：上述代码中的阈值、轮廓查找方式等参数需要根据具体情况进行调整。