Raspberry Pi ロボットカー OpenCV(画像認識)

2019年に製作したRaspberry Pi ロボットカー OpenCV(画像認識)

について記事にしました。

Mean-Shift 法を用いた移動物体追跡による制御を行いました。 

同ブログ関連記事：【Raspberry Pi ロボットカー回路図】

 

動画

画像認識

ピンク色のふたを追跡します。

OpenCV インストール

$ sudo apt-get update ; sudo apt-get upgrade   #システム環境をアップデートしておく

$ sudo apt-get install build-essential cmake cmake-qt-gui  #cmakeをインストール

$ sudo apt-get install build-essential git cmake pkg-config libjpeg-dev libtiff5-dev libjasper-dev libpng12-dev libavcodec-dev libavformat-dev libswscale-dev libv4l-dev

libgtk2.0-dev libatlas-base-dev gfortran  #必要なパッケージをインストール

$ sudo apt-get install python3-dev python3-pip  #Python3関連のセットアップ

$ cd ~

$ git clone https://github.com/Itseez/opencv.git

$ cd opencv

$ git checkout 3.2.0

$ cd 

$ sudo pip3 install numpy  #numpyのインストール

$ cd opencv

$ mkdir build

$ cd build

$ cmake -D CMAKE_BUILD_TYPE=RELEASE \  #OpenCVのインストール

>     -D CMAKE_INSTALL_PREFIX=/usr/local \

>     -D INSTALL_PYTHON_EXAMPLES=ON \

>     -D OPENCV_EXTRA_MODULES_PATH=~/opencv_contrib/modules \

>     -D BUILD_EXAMPLES=ON ..

#　Remove previous version

$ sudo apt autoremove libopencv3

# Install 

wget https://github.com/mt08xx/files/raw/master/opencv-rpi/libopencv3_3.4.0-20180115.1_armhf.deb

$ sudo apt install -y ./libopencv3_3.4.0-20180115.1_armhf.deb

$ sudo ldconfig

$ cd opencv

$ mkdir build

$ cd build

$ cmake -DCMAKE_BUILD_TYPE=Release \

>    -D CMAKE_INSTALL_PREFIX=/usr/local \

>    -D OPENCV_EXTRA_MODULES_PATH=/home/npi/opencv_contrib-3.3.1/modules \

>    -DENABLE_VFPV3=ON \

>    -DENABLE_NEON=ON \

>    -DBUILD_TESTS=OFF \

>    -DWITH_TBB=OFF \

>    -D INSTALL_PYTHON_EXAMPLES=ON \

>    -D BUILD_EXAMPLES=ON ..

プログラム（meanshift_car.py）

# -*- coding: utf-8 -*-

import smbus

import numpy as np

import cv2

import time

import Adafruit_PCA9685

import RPi.GPIO as GPIO

import Adafruit_GPIO.SPI as SPI

import Adafruit_SSD1306 

from PIL import Image

from PIL import ImageDraw

from PIL import ImageFont 

disp = Adafruit_SSD1306.SSD1306_128_64(rst=0)

FONT = '/usr/share/fonts/truetype/fonts-japanese-gothic.ttf' 

disp.begin()

disp.clear()

disp.display()

image = Image.new('1', (128, 64) ) 

draw = ImageDraw.Draw(image) 

size1 = ImageFont.truetype(FONT, 20 )

image = Image.open('robot5.ppm').convert('1')

disp.image(image)

disp.display()

i2c = smbus.SMBus(1)

i2c.write_byte_data(0x3c ,0x00 ,0x2f)#スクロールon

GPIO.setmode(GPIO.BCM) 

GPIO.setup(5, GPIO.OUT)      #AIN1

GPIO.setup(6, GPIO.OUT)      #BIN1

GPIO.setup(12, GPIO.OUT)   #AIN2

GPIO.setup(13, GPIO.OUT)   #BIN2

AIN1= GPIO.PWM(5,100)            #GPIO05 as PWM output, with 100Hz frequency

BIN1= GPIO.PWM(6,100)            #GPIO06 as PWM output, with 100Hz frequency 

AIN2 = GPIO.PWM(12,100)         #GPIO12 as PWM output, with 100Hz frequency

BIN2= GPIO.PWM(13,100)           #GPIO13 as PWM output, with 100Hz frequency 

AIN1.start(0)

BIN1.start(0)

AIN2.start(0)

BIN2.start(0)

def car_s():

AIN1.ChangeDutyCycle(0)

BIN1.ChangeDutyCycle(0)

AIN2.ChangeDutyCycle(0)

BIN2.ChangeDutyCycle(0)

def car_f():

AIN1.ChangeDutyCycle(10)

BIN1.ChangeDutyCycle(10)

AIN2.ChangeDutyCycle(0)

BIN2.ChangeDutyCycle(0)

def car_r():

AIN1.ChangeDutyCycle(10)

BIN1.ChangeDutyCycle(0)

AIN2.ChangeDutyCycle(0)

BIN2.ChangeDutyCycle(10)

def car_l():

AIN1.ChangeDutyCycle(0)

BIN1.ChangeDutyCycle(10)

AIN2.ChangeDutyCycle(10)

BIN2.ChangeDutyCycle(0)

def car_b():

AIN1.ChangeDutyCycle(0)

BIN1.ChangeDutyCycle(0)

AIN2.ChangeDutyCycle(10)

BIN2.ChangeDutyCycle(10)

pwm = Adafruit_PCA9685.PCA9685()

pwm.set_pwm_freq(60)

cap = cv2.VideoCapture(0)

cap.set(3, 320)

cap.set(4, 320)

x, y, w, h = 75, 75, 100, 100

track_window = (x, y, w, h)

ret,frame = cap.read()

roi = frame[y:y+h, x:x+w]

hsv_roi =  cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)

img_mask = cv2.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))

roi_hist = cv2.calcHist([hsv_roi], [0], img_mask, [180], [0,180])

cv2.normalize(roi_hist, roi_hist, 0, 255, cv2.NORM_MINMAX)

term_crit = ( cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 10, 1 )

while(True):

    ret, frame = cap.read()

    now_degree_x, now_degree_y, move_degree_x, move_degree_y = 300, 250, 0, 0

    if ret == True:

        hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)

        dst = cv2.calcBackProject([hsv],[0],roi_hist,[0,180], 1)

        ret, track_window = cv2.meanShift(dst, track_window, term_crit)

        x,y,w,h = track_window

        img_dst = cv2.rectangle(frame, (x,y), (x+w, y+h), (255,255,255), 3)

        img_x = int(x+w/2)

        img_y = int(y+h/2)

        move_degree_x = now_degree_x - (img_x-160)*0.06

        move_degree_y = now_degree_y + (img_y-160)*0.06

        print('deg: ', move_degree_x , move_degree_y)

        pwm.set_pwm(0, 0, int(move_degree_x))

        pwm.set_pwm(1, 0, int(move_degree_y)) 

        if move_degree_x < 292:

car_r()

time.sleep(0.5)

car_s()

        if move_degree_x > 306:

car_l()

time.sleep(0.5)

car_s()

        if move_degree_y <246:

car_b()

time.sleep(0.5)

car_s()

        if move_degree_y >= 254:

car_f()

time.sleep(0.5)

car_s()

       

       

        now_degree_x = now_degree_x + move_degree_x

        now_degree_y = now_degree_y + move_degree_y

        cv2.circle(img_dst, (img_x, img_y), 10, (0,180,0), -1)

        cv2.imshow('SHOW MEANSHIFT IMAGE', img_dst)

       

        k = cv2.waitKey(1)

        if k == ord('q'):

            break

    else:

        break

cap.release()

cv2.destroyAllWindows()