Python OpenCV 边缘检测

# -*- coding:utf-8 -*-
import cv2
import numpy as np

'''
  步骤：
    (1).彩色图像转换为灰度图像（以灰度图或者单通道图读入）
    (2).对图像进行高斯模糊（去噪）
    (3).计算图像梯度，根据梯度计算图像边缘幅值与角度
    (4).沿梯度方向进行非极大值抑制（边缘细化）
    (5).双阈值边缘连接处理
    (6).二值化图像输出结果
'''

cv2.namedWindow('Cannys', 0)
# 创建滑动条
cv2.createTrackbar('minval', 'Cannys', 120, 300, lambda x: None)
cv2.createTrackbar('maxval', 'Cannys', 200, 300, lambda x: None)
cv2.createTrackbar('blur', 'Cannys', 12, 100, lambda x: None)


# 绘制等高线轮廓
def draw_Contour_Line(frame):
    # 去噪
    blur = cv2.getTrackbarPos('blur', 'Cannys')
    frame = cv2.GaussianBlur(frame, (5, 5), blur * 0.1)
    # 读取滑动条数值
    minval = cv2.getTrackbarPos('minval', 'Cannys')
    maxval = cv2.getTrackbarPos('maxval', 'Cannys')
    # threshold1 threshold2 两个阈值，小的控制边缘连接，大的控制强边缘的初始分割。如果一个像素的梯度大于上限值，则认为是边缘像素，如果小于下限阈值，则抛弃，如若点的梯度在两者之间，则当这个点与高于上限值的像素点连接时才保留，否则删除。
    # aperture_size 算子内核大小，表示Sobel 算子大小，默认为3即表示一个3*3的矩阵
    canny = cv2.Canny(frame, threshold1=minval, threshold2=maxval)
    # canny = cv2.Canny(frame, threshold1=60, threshold2=180)
    # RETR_EXTERNAL:表示只检测最外层轮廓;  RETR_CCOMP:提取所有轮廓;  RETR_TREE:提取所有轮廓并重新建立网状轮廓结构
    # CHAIN_APPROX_SIMPLE：压缩水平方向，垂直方向，对角线方向的元素，值保留该方向的重点坐标;  CHAIN_APPROX_NONE：获取每个轮廓的每个像素，相邻的两个点的像素位置差不超过1
    contours, hierarchy = cv2.findContours(canny, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    frameProcessed = np.zeros(frame.shape, dtype=np.uint8)
    frameProcessed = cv2.cvtColor(frameProcessed, cv2.COLOR_GRAY2BGR)
    cv2.drawContours(frameProcessed, contours, -1, color=(255, 0, 0), thickness=2)  # blue
    return frameProcessed


def draw_Contour_Line2(frame):
    frame = cv2.GaussianBlur(frame, (5, 5), 1.5)
    minval = cv2.getTrackbarPos('minval', 'Cannys')
    maxval = cv2.getTrackbarPos('maxval', 'Cannys')
    canny = cv2.Canny(frame, threshold1=minval, threshold2=maxval)
    # canny = cv2.Canny(frame, 60, 180)
    return canny


def draw_Contour_Line3(frame):
    frame = cv2.GaussianBlur(frame, (5, 5), 1.5)
    canny = cv2.Canny(frame, threshold1=60, threshold2=180)
    # 形态学：边缘检测
    _, Thr_img = cv2.threshold(canny, 210, 255, cv2.THRESH_BINARY)  # 设定红色通道阈值210（阈值影响梯度运算效果）
    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))  # 定义矩形结构元素
    gradient = cv2.morphologyEx(Thr_img, cv2.MORPH_GRADIENT, kernel)  # 梯度

    return gradient


camera = cv2.VideoCapture(0)
while True:
    ret, frame = camera.read()
    gray_L = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    frame_edge1 = draw_Contour_Line(gray_L)
    frame_edge2 = draw_Contour_Line2(gray_L)
    frame_edge3 = draw_Contour_Line3(gray_L)
    cv2.imshow("frame", frame)
    cv2.imshow("Canny-contour1", frame_edge1)
    cv2.imshow("Canny-contour2", frame_edge2)
    cv2.imshow("Canny-contour3", frame_edge3)
    if cv2.waitKey(1) == ord("q"):
        break

camera.release()
cv2.destroyAllWindows()