目錄

試錯(cuò)

試錯(cuò)1：形態(tài)學(xué)處理

試錯(cuò)2：HSV色彩空間

基礎(chǔ)理論

1、HSV與HSL色彩空間

2、PID調(diào)節(jié)

一、OpenCV圖像處理

1、在HSL色彩空間下得到二值圖

2、 對二值圖形態(tài)學(xué)處理

3、找出線的輪廓和中心點(diǎn)坐標(biāo)

二、PID

三、運(yùn)動(dòng)控制

總代碼

---

試錯(cuò)

試錯(cuò)1：形態(tài)學(xué)處理

一開始用的形態(tài)學(xué)處理，自行改變閾值，調(diào)試之后，進(jìn)行處理，發(fā)現(xiàn)效果不是太好，于是改成了HSV色彩空間。

試錯(cuò)2：HSV色彩空間

之前沒注意到，HSV色彩空間很難識(shí)別白色：

HSV：?

?不難看出，如果尋白色線的話，HSV色彩空間不是一個(gè)很好的選擇，下面引入HSL色彩空間：

?HSL：

所以，如果是巡白色的話，建議用HSL色彩空間。

注意：巡線小車的攝像頭不能太低，如果太低了，可能讓小車自己的影子會(huì)阻礙光線。

hsv中的效果：

?

hsl中的效果：

可以看出，已經(jīng)能大致找到白線了。

基礎(chǔ)理論

1、HSV與HSL色彩空間

?HSV：?

?不難看出，如果尋白色線的話，HSV色彩空間不是一個(gè)很好的選擇，下面引入HSL色彩空間：

?HSL：

所以，如果是巡白色的話，建議用HSL色彩空間。

2、PID調(diào)節(jié)

個(gè)人理解：

P：拉力

I：推動(dòng)力

D：阻力?

?

?

一、OpenCV圖像處理

?

1、在HSL色彩空間下得到二值圖

?

?

?

?

# 在HSV色彩空間下得到二值圖
def Get_HSV(image):# 1 get trackbar's valuehmin = cv2.getTrackbarPos('hmin', 'h_binary')hmax = cv2.getTrackbarPos('hmax', 'h_binary')smin = cv2.getTrackbarPos('smin', 's_binary')smax = cv2.getTrackbarPos('smax', 's_binary')lmin = cv2.getTrackbarPos('lmin', 'l_binary')lmax = cv2.getTrackbarPos('lmax', 'l_binary')# 2 to HSVhls = cv2.cvtColor(image, cv2.COLOR_BGR2HLS)cv2.imshow('hls', hls)h, l, s = cv2.split(hls)# 3 set threshold (binary image)# if value in (min, max):white; otherwise:blackh_binary = cv2.inRange(np.array(h), np.array(hmin), np.array(hmax))s_binary = cv2.inRange(np.array(s), np.array(smin), np.array(smax))l_binary = cv2.inRange(np.array(l), np.array(lmin), np.array(lmax))# 4 get binary（對H、S、V三個(gè)通道分別與操作）binary = 255 - cv2.bitwise_and(h_binary, cv2.bitwise_and(s_binary, l_binary))# 5 Showcv2.imshow('h_binary', h_binary)cv2.imshow('s_binary', s_binary)cv2.imshow('l_binary', l_binary)cv2.imshow('binary', binary)return binary

2、 對二值圖形態(tài)學(xué)處理

?

?

?

# 圖像處理
def Image_Processing():global frame, binary# Capture the framesret, frame = camera.read()# to binarybinary = Get_HSV(frame)blur = cv2.GaussianBlur(binary, (5, 5), 0)cv2.imshow('blur', blur)kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (35, 35))Open = cv2.morphologyEx(blur, cv2.MORPH_OPEN, kernel)cv2.imshow('Open', Open)kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (25, 25))Erode = cv2.morphologyEx(Open, cv2.MORPH_ERODE, kernel)cv2.imshow('Erode', Erode)kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (25, 25))Dilate = cv2.morphologyEx(Erode, cv2.MORPH_DILATE, kernel)cv2.imshow('Dilate', Dilate)binary = Erode#Dilate

3、找出線的輪廓和中心點(diǎn)坐標(biāo)

?

?

# 找線
def Find_Line():global x, y, image# 1 找出所有輪廓bin2, contours, hierarchy = cv2.findContours(binary, 1, cv2.CHAIN_APPROX_NONE)# 2 找出最大輪廓if len(contours) > 0:# 最大輪廓c = max(contours, key=cv2.contourArea)M = cv2.moments(c)# 中心點(diǎn)坐標(biāo)x = int(M['m10'] / M['m00'])y = int(M['m01'] / M['m00'])#print(x, y)# 顯示image = frame.copy()# 標(biāo)出中心位置cv2.line(image, (x, 0), (x, 720), (0, 0, 255), 1)cv2.line(image, (0, y), (1280, y), (0, 0, 255), 1)# 畫出輪廓cv2.drawContours(image, contours, -1, (128, 0, 128), 2)cv2.imshow("image", image)else:print("not found the line")(x,y) = (0, 0)

?

二、PID

比例：獲取當(dāng)前時(shí)刻白線中心點(diǎn)與圖像中點(diǎn)的誤差，作為當(dāng)前誤差。

積分：獲取上一時(shí)刻的誤差。

def Pid():global turn_speed, x, y, speedglobal error, last_error, pre_error, out_piderror = abs(x - width / 2)out_pid = int(proportion * error - integral * last_error + derivative * pre_error)turn_speed = out_pid# 保存本次誤差，以便下一次運(yùn)算pre_error = last_errorlast_error = error# 限值if (turn_speed < 30):turn_speed = 30elif (turn_speed > 100):turn_speed = 100if (speed < 0):speed = 0elif (speed > 100):speed = 100print(error, out_pid, turn_speed, (x, y))

三、運(yùn)動(dòng)控制

# 巡線
def Follow_Line():global turn_speed, x, y,speed, back_speed'''if(x < width / 2 and y>2*height/3):Left(turn_speed)elif(x>3*width/2 and y>2*height/3):Right(turn_speed)'''if(0<x<width/4):Left(turn_speed)print("turn left")elif(3*width/4<x<width):Right(turn_speed)print("turn right")#直角拐彎elif(y>3*height/4):if(x<width/2):Left(turn_speed*2)print("turn left")elif(x>=width/2):Right(turn_speed*2)print("turn right")elif(x>=width/4 and x<=3*width/4):Forward(speed)elif(x==0 and y==0):Back(back_speed)

總代碼

#!/usr/bin/env python2
# -*- coding: utf-8 -*-import numpy as np
import cv2
import Adafruit_PCA9685
import RPi.GPIO as GPIO
import timel_motor = 18
left_Forward = 22
left_back = 27r_motor = 23
right_Forward = 25
right_back = 24pwm_servo = Adafruit_PCA9685.PCA9685()width, height = 160, 120
camera = cv2.VideoCapture(0)
camera.set(3, width)
camera.set(4, height)# pid
error = 0  # 當(dāng)前誤差e[k]
last_error = 0  # 上一次誤差e[k-1]
pre_error = 0  # 上上次誤差e[k-2]
proportion = 1  # 比例系數(shù)3 0.2
integral = 0.5  # 積分系數(shù)1.2
derivative = 0  # 微分系數(shù)1.2stop_flag = 1
control_flag = 1
turn_speed = 30
speed = 30
back_speed = 30def Motor_Init():global L_Motor, R_MotorL_Motor = GPIO.PWM(l_motor, 100)R_Motor = GPIO.PWM(r_motor, 100)L_Motor.start(0)R_Motor.start(0)def Direction_Init():GPIO.setup(left_back, GPIO.OUT)GPIO.setup(left_Forward, GPIO.OUT)GPIO.setup(l_motor, GPIO.OUT)GPIO.setup(right_Forward, GPIO.OUT)GPIO.setup(right_back, GPIO.OUT)GPIO.setup(r_motor, GPIO.OUT)def set_servo_angle(channel, angle):angle = 4096 * ((angle * 11) + 500) / 20000pwm_servo.set_pwm_freq(50)  # frequency==50Hz (servo)pwm_servo.set_pwm(channel, 0, int(angle))def TrackBar_Init():# 1 create windowscv2.namedWindow('h_binary')cv2.namedWindow('s_binary')cv2.namedWindow('l_binary')# 2 Create Trackbarcv2.createTrackbar('hmin', 'h_binary', 0, 179, call_back)cv2.createTrackbar('hmax', 'h_binary', 110, 179, call_back)cv2.createTrackbar('smin', 's_binary', 0, 255, call_back)cv2.createTrackbar('smax', 's_binary', 51, 255, call_back)  # 51cv2.createTrackbar('lmin', 'l_binary', 0, 255, call_back)cv2.createTrackbar('lmax', 'l_binary', 255, 255, call_back)'''cv2.namedWindow('binary')cv2.createTrackbar('thresh', 'binary', 154, 255, call_back)  '''#   創(chuàng)建滑動(dòng)條     滑動(dòng)條值名稱 窗口名稱   滑動(dòng)條值 滑動(dòng)條閾值 回調(diào)函數(shù)def Init():GPIO.setwarnings(False)GPIO.setmode(GPIO.BCM)Direction_Init()Motor_Init()TrackBar_Init()def Forward(turn_speed):L_Motor.ChangeDutyCycle(turn_speed)GPIO.output(left_Forward, 1)  # left_ForwardGPIO.output(left_back, 0)  # left_backR_Motor.ChangeDutyCycle(turn_speed)GPIO.output(right_Forward, 1)  # right_ForwardGPIO.output(right_back, 0)  # right_backdef Back(turn_speed):L_Motor.ChangeDutyCycle(turn_speed)GPIO.output(left_Forward, 0)  # left_ForwardGPIO.output(left_back, 1)  # left_backR_Motor.ChangeDutyCycle(turn_speed)GPIO.output(right_Forward, 0)  # right_ForwardGPIO.output(right_back, 1)  # right_backdef Left(turn_speed):L_Motor.ChangeDutyCycle(turn_speed)GPIO.output(left_Forward, 0)  # left_ForwardGPIO.output(left_back, 1)  # left_backR_Motor.ChangeDutyCycle(turn_speed)GPIO.output(right_Forward, 1)  # right_ForwardGPIO.output(right_back, 0)  # right_backdef Right(turn_speed):L_Motor.ChangeDutyCycle(turn_speed)GPIO.output(left_Forward, 1)  # left_ForwardGPIO.output(left_back, 0)  # left_backR_Motor.ChangeDutyCycle(turn_speed)GPIO.output(right_Forward, 0)  # right_ForwardGPIO.output(right_back, 1)  # right_backdef Stop():L_Motor.ChangeDutyCycle(0)GPIO.output(left_Forward, 0)  # left_ForwardGPIO.output(left_back, 0)  # left_backR_Motor.ChangeDutyCycle(0)GPIO.output(right_Forward, 0)  # right_ForwardGPIO.output(right_back, 0)  # right_back# 回調(diào)函數(shù)
def call_back(*arg):pass# 在HSV色彩空間下得到二值圖
def Get_HSV(image):# 1 get trackbar's valuehmin = cv2.getTrackbarPos('hmin', 'h_binary')hmax = cv2.getTrackbarPos('hmax', 'h_binary')smin = cv2.getTrackbarPos('smin', 's_binary')smax = cv2.getTrackbarPos('smax', 's_binary')lmin = cv2.getTrackbarPos('lmin', 'l_binary')lmax = cv2.getTrackbarPos('lmax', 'l_binary')# 2 to HSVhls = cv2.cvtColor(image, cv2.COLOR_BGR2HLS)cv2.imshow('hls', hls)h, l, s = cv2.split(hls)# 3 set threshold (binary image)# if value in (min, max):white; otherwise:blackh_binary = cv2.inRange(np.array(h), np.array(hmin), np.array(hmax))s_binary = cv2.inRange(np.array(s), np.array(smin), np.array(smax))l_binary = cv2.inRange(np.array(l), np.array(lmin), np.array(lmax))# 4 get binary（對H、S、V三個(gè)通道分別與操作）binary = 255 - cv2.bitwise_and(h_binary, cv2.bitwise_and(s_binary, l_binary))# 5 Showcv2.imshow('h_binary', h_binary)cv2.imshow('s_binary', s_binary)cv2.imshow('l_binary', l_binary)cv2.imshow('binary', binary)return binary# 手動(dòng)控制小車（上下左右，案件事件判斷）
# 控制方式：w、s、a、d分別表示：上、下、左、右
def Key_Control(keyboard):global stop_flag, control_flagif keyboard == ord("w"):Forward(50)time.sleep(0.1)Stop()elif keyboard == ord("s"):Back(50)time.sleep(0.1)Stop()elif keyboard == ord("a"):Left(50)time.sleep(0.1)Stop()elif keyboard == ord("d"):Right(50)time.sleep(0.1)Stop()# 圖像處理
def Image_Processing():global frame, binary# Capture the framesret, frame = camera.read()# to binarybinary = Get_HSV(frame)blur = cv2.GaussianBlur(binary, (5, 5), 0)cv2.imshow('blur', blur)kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (35, 35))Open = cv2.morphologyEx(blur, cv2.MORPH_OPEN, kernel)cv2.imshow('Open', Open)kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (25, 25))Erode = cv2.morphologyEx(Open, cv2.MORPH_ERODE, kernel)cv2.imshow('Erode', Erode)kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (25, 25))Dilate = cv2.morphologyEx(Erode, cv2.MORPH_DILATE, kernel)cv2.imshow('Dilate', Dilate)binary = Erode  # Dilate# 找線
def Find_Line():global x, y, image# 1 找出所有輪廓bin2, contours, hierarchy = cv2.findContours(binary, 1, cv2.CHAIN_APPROX_NONE)# 2 找出最大輪廓if len(contours) > 0:# 最大輪廓c = max(contours, key=cv2.contourArea)M = cv2.moments(c)# 中心點(diǎn)坐標(biāo)x = int(M['m10'] / M['m00'])y = int(M['m01'] / M['m00'])# print(x, y)# 顯示image = frame.copy()# 標(biāo)出中心位置cv2.line(image, (x, 0), (x, 720), (0, 0, 255), 1)cv2.line(image, (0, y), (1280, y), (0, 0, 255), 1)# 畫出輪廓cv2.drawContours(image, contours, -1, (128, 0, 128), 2)cv2.imshow("image", image)else:print("not found the line")(x, y) = (0, 0)def Pid():global turn_speed, x, y, speedglobal error, last_error, pre_error, out_piderror = abs(x - width / 2)out_pid = int(proportion * error - integral * last_error + derivative * pre_error)turn_speed = out_pid# 保存本次誤差，以便下一次運(yùn)算pre_error = last_errorlast_error = error# 限值if (turn_speed < 30):turn_speed = 30elif (turn_speed > 100):turn_speed = 100if (speed < 0):speed = 0elif (speed > 100):speed = 100print(error, out_pid, turn_speed, (x, y))# 巡線
def Follow_Line():global turn_speed, x, y, speed, back_speed'''if(x < width / 2 and y>2*height/3):Left(turn_speed)elif(x>3*width/2 and y>2*height/3):Right(turn_speed)'''if (0 < x < width / 4):Left(turn_speed)print("turn left")elif (3 * width / 4 < x < width):Right(turn_speed)print("turn right")# 直角拐彎elif (y > 3 * height / 4):if (x < width / 2):Left(turn_speed * 2)print("turn left")elif (x >= width / 2):Right(turn_speed * 2)print("turn right")elif (x >= width / 4 and x <= 3 * width / 4):Forward(speed)elif (x == 0 and y == 0):Back(back_speed)def Control():global control_flag, speed, proportion, integralkeyboard = cv2.waitKey(1)# 加速減速if (keyboard == ord('k')):speed += 5elif (keyboard == ord('l')):speed -= 5print(speed)if keyboard == ord("n"):integral += 0.01elif keyboard == ord("m"):integral -= 0.01print(integral)if (control_flag == -1):Follow_Line()if keyboard == 32:control_flag *= -1Stop()else:Key_Control(keyboard)if keyboard == 32:control_flag *= -1Stop()print(control_flag)if __name__ == '__main__':Init()set_servo_angle(4, 140)  # top servo     lengthwise# 0:back    180:frontset_servo_angle(5, 90)  # bottom servo  crosswise# 0:left    180:rightwhile True:Image_Processing()Find_Line()Pid()Control()if cv2.waitKey(1) == ord('q'):cv2.destroyAllWindows()break

?其實(shí)一開始主要是想玩機(jī)器視覺，小車的運(yùn)動(dòng)控制研究的不算精細(xì)，PID研究的也不深。

有很多是自己的想法，有錯(cuò)誤歡迎指正，有建議也歡迎交流，謝謝。

總結(jié)

以上是生活随笔為你收集整理的树莓派视觉小车 -- OpenCV巡线(HSL色彩空间、PID)的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。