Computer vision 7 (모폴로지 연산과 객체 검출)

2023-03-28 6 분 소요

모폴로지 연산과 객체 검출

1. 모폴로지 연산

침식과 팽창

import cv2
import numpy as np

src = cv2.imread('./data/milkdrop.bmp', cv2.IMREAD_GRAYSCALE)
thresh, src_bin = cv2.threshold(src, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
print(thresh)

erode = cv2.erode(src_bin, None)
dilate = cv2.dilate(src_bin, None)

cv2.imshow('src', src)
cv2.imshow('src_bin', src_bin)
cv2.imshow('erode', erode)
cv2.imshow('dilate', dilate)

cv2.waitKey()
cv2.destroyAllWindows()

154.0

열기와 닫기

src = cv2.imread('./data/milkdrop.bmp', cv2.IMREAD_GRAYSCALE)
thresh, src_bin = cv2.threshold(src, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
print(thresh)

# cv2.morphologyEx(원본영상, 연산방법, 구조물, 반복횟수)

# 열기 : 침식 -> 팽창
opening = cv2.morphologyEx(src_bin, cv2.MORPH_OPEN, None, iterations=1)

# 닫기 : 팽창 -> 침식
closing = cv2.morphologyEx(src_bin, cv2.MORPH_CLOSE, None, iterations=1)

cv2.imshow('src', src)
cv2.imshow('src_bin', src_bin)
cv2.imshow('opening', opening)
cv2.imshow('closing', closing)

cv2.waitKey()
cv2.destroyAllWindows()

154.0

2. 레이블링과 외곽선 검출

레이블링 기본

src = np.array([[0, 0, 1, 1, 0, 0, 0, 0],
                [1, 1, 1, 1, 0, 0, 1, 0],
                [1, 1, 1, 1, 0, 0, 0, 0],
                [0, 0, 0, 0, 0, 1, 1, 0],
                [0, 0, 0, 1, 1, 1, 1, 0],
                [0, 0, 0, 1, 0, 0, 1, 0],
                [0, 0, 1, 1, 1, 1, 1, 0],
                [0, 0, 0, 0, 0, 0, 0, 0]]).astype(np.uint8)

src = src * 255

cnt, labels = cv2.connectedComponents(src)

cnt # 4개의 그룹(3개 객체 + 1개 배경)

labels

array([[0, 0, 1, 1, 0, 0, 0, 0],
       [1, 1, 1, 1, 0, 0, 2, 0],
       [1, 1, 1, 1, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 3, 3, 0],
       [0, 0, 0, 3, 3, 3, 3, 0],
       [0, 0, 0, 3, 0, 0, 3, 0],
       [0, 0, 3, 3, 3, 3, 3, 0],
       [0, 0, 0, 0, 0, 0, 0, 0]], dtype=int32)

src = cv2.imread('./data/circles.jpg', cv2.IMREAD_GRAYSCALE)
thresh, src_bin = cv2.threshold(src, 0, 255, cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)

# todo 객체별로 색깔 다르게 해서 보여주기
cnt, labels = cv2.connectedComponents(src_bin)

dst = cv2.cvtColor(src, cv2.COLOR_GRAY2BGR) # 객체별로 색깔을 다르게 표시할 3차원 도화지 준비

dst[labels==0] = (0, 255, 255) # 배경을 노란색을 설정
dst[labels==1] = (0, 0, 255) # 1번객체를  빨간색으로 설정
dst[labels==2] = (255, 0, 255) # 2번객체를 보라색으로 설정
dst[labels==3] = (0, 255, 0) #3번 객체를 초록색으로 설정

cv2.imshow('src', src)
cv2.imshow('src_bin', src_bin)
cv2.imshow('dst', dst)

cv2.waitKey()
cv2.destroyAllWindows()

레이블링 응용

src = cv2.imread('./data/circles.jpg', cv2.IMREAD_GRAYSCALE)
thresh, src_bin = cv2.threshold(src, 0, 255, cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)


cnt, labels, stats, centroids = cv2.connectedComponentsWithStats(src_bin)

dst = cv2.cvtColor(src, cv2.COLOR_GRAY2BGR) # 객체별로 색깔을 다르게 표시할 3차원 도화지 준비

for i in range(1, cnt): # 객체 1, 2, 3 (배경 제외)
    b = np.random.randint(0, 255) 
    g = np.random.randint(0, 255) 
    r = np.random.randint(0, 255) 
    
    dst[labels==i] = (b, g, r)
    
    # boungding box (from stats) --> rectangle 그리기
    x, y, width, height, area = stats[i]
    cv2.rectangle(dst, (x, y), (x+width, y+height), (0, 0, 255))
    
    # center (from centroids) --> circle 그리기
    cx, cy = centroids[i]
    cv2.circle(dst, (int(cx), int(cy)), 5, (255, 0, 0), -1)

cv2.imshow('src', src)
cv2.imshow('src_bin', src_bin)
cv2.imshow('dst', dst)

cv2.waitKey()
cv2.destroyAllWindows()

src = cv2.imread('./data/keyboard.bmp')
dst = src.copy()

gray = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
thresh, src_bin = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)

cnt, labels, stats, centroids = cv2.connectedComponentsWithStats(src_bin)

for i in range(1, cnt): # 객체 1~37 (배경 제외)   
    # boungding box (from stats) --> rectangle 그리기
    x, y, width, height, area = stats[i]
    if area > 20:
        cv2.rectangle(dst, (x, y), (x+width, y+height), (0, 255, 255))

cv2.imshow('src', src)
cv2.imshow('src_bin', src_bin)
cv2.imshow('dst', dst)
cv2.waitKey()
cv2.destroyAllWindows()

외곽선 검출과 그리기

src = cv2.imread('./data/contours.bmp')
dst = src.copy()

gray = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)

# thresh, src_bin = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY)

contours, hierarchy = cv2.findContours(gray, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)
for i in range(len(contours)): # 9 iterations
    b = np.random.randint(0, 255) 
    g = np.random.randint(0, 255) 
    r = np.random.randint(0, 255)
    cv2.drawContours(dst, contours, i, (b, g, r), 2)

cv2.imshow('src', src)
cv2.imshow('gray', gray)
cv2.imshow('dst', dst)
cv2.waitKey()
cv2.destroyAllWindows()

실습 thumbs_up_down.jpg의 손 외곽선 그리기

src = cv2.imread('./data/thumbs_up_down.jpg')
dst = src.copy()

gray = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
thresh, gray_bin = cv2.threshold(gray, 220, 255, cv2.THRESH_BINARY_INV)
print(thresh)

contours, hierarchy = cv2.findContours(gray_bin, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)
cv2.drawContours(dst, contours, -1, (0, 0, 255), 3, cv2.LINE_AA)

cv2.imshow('src', src)
cv2.imshow('gray', gray)
cv2.imshow('gray_bin', gray_bin)
cv2.imshow('dst', dst)
cv2.waitKey()
cv2.destroyAllWindows()

220.0

# 오른손 왼손을 둘러싸는 Bounding Box 그리기

right_hand = contours[0] # 오른손을 둘러싸는 점의 좌표들
left_hand = contours[1] # 왼손을 둘러싸는 점의 좌표들
x1, y1, width1, height1 = cv2.boundingRect(right_hand)
x2, y2, width2, height2 = cv2.boundingRect(left_hand)

dst = src.copy()
cv2.rectangle(dst, (x1, y1), (x1+width1, y1+height1), (0, 255, 255), 3)
cv2.rectangle(dst, (x2, y2), (x2+width2, y2+height2), (0, 0, 255), 3)

cv2.imshow('dst', dst)
cv2.waitKey()
cv2.destroyAllWindows()

3.객체 검출

캐스케이드 분류기와 얼굴 검출

https://docs.opencv.org/4.6.0/db/d28/tutorial_cascade_classifier.html
https://towardsdatascience.com/viola-jones-algorithm-and-haar-cascade-classifier-ee3bfb19f7d8

얼굴 검출

image = cv2.imread('./data/lena.jpg')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

face_cascade = cv2.CascadeClassifier('./data/haarcascade_frontalface_default.xml')
faces = face_cascade.detectMultiScale(gray)

for face in faces:
    x, y, w, h = face
    cv2.rectangle(image, (x, y), (x+w, y+h), (0, 0, 255), 2)
    
cv2.imshow('image', image)    
cv2.waitKey()
cv2.destroyAllWindows()

눈 검출

image = cv2.imread('./data/kids2.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

face_cascade = cv2.CascadeClassifier('./data/haarcascade_frontalface_default.xml')
eye_cascade = cv2.CascadeClassifier('./data/haarcascade_eye.xml')

faces = face_cascade.detectMultiScale(gray)

for face in faces:
    x, y, w, h = face
    cv2.rectangle(image, (x, y), (x+w, y+h), (0, 0, 255), 2)
    face_rect = image[y:y+h, x:x+w]
    eyes = eye_cascade.detectMultiScale(face_rect)
    for eye in eyes:
        x2, y2, w2, h2 = eye
        cv2.rectangle(face_rect, (x2, y2), (x2+w2, y2+h2), (255, 0, 0), 2)    
    
cv2.imshow('image', image)    
cv2.waitKey()
cv2.destroyAllWindows()

# detecMultiScale() 의 중요 파라미터들
# scaleFactor : 검색 윈도우 확대 비율(default = 1.1)
# minNeighbors : 검출 영역으로 선택하기 위한 최소 검출 횟수 (default=3)
# minSize : 검출할 객체의 최소 크기 (예, (100, 100))
# maxSize : 검출할 객체의 최대 크기

# scaleFactor가 detect시 미치는 영향
# https://answers.opencv.org/question/10654/how-does-the-parameter-scalefactor-in-detectmultiscale-affect-face-detection/

실습 : multi face 검출

image = cv2.imread('./data/multi_faces.jpg')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

face_cascade = cv2.CascadeClassifier('./data/haarcascade_frontalface_default.xml')
faces = face_cascade.detectMultiScale(gray, scaleFactor=1.3, minNeighbors=8)
#faces = face_cascade.detectMultiScale(gray, minSize=(90, 90), maxSize=(100, 100))
#faces = face_cascade.detectMultiScale(gray, minNeighbors=15)

for face in faces:
    x, y, w, h = face
    cv2.rectangle(image, (x, y), (x+w, y+h), (0, 0, 255), 2)

cv2.imshow('image', image)    
cv2.waitKey()
cv2.destroyAllWindows()

실습 : Camera Device로부터 얻은 frame에 대해 얼굴(눈) 검출하기

import sys

face_cascade = cv2.CascadeClassifier('./data/haarcascade_frontalface_default.xml')
eye_cascade = cv2.CascadeClassifier('./data/haarcascade_eye.xml')

cap = cv2.VideoCapture(0) # 0: 메인 카메라

if not cap.isOpened():
    print("camera open failed")
    sys.exit()
    
width = cap.get(cv2.CAP_PROP_FRAME_WIDTH)    
height = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)    
fps = cap.get(cv2.CAP_PROP_FPS) # frame per second
delay = round(1000/30)

print(width, height, fps, delay)

while True:    
    ret, frame = cap.read() # frame : 이미지 한장 (shape : height x width x channel)        
    
    if not ret:
        print("frame read error")
        break
        
    # detect face and eyes
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray, minNeighbors=10)
    
    for face in faces:
        x1, y1, w1, h1 = face
        cv2.rectangle(frame, (x1, y1), (x1+w1, y1+h1), (0, 0, 255), 2)
        face_rect = frame[y1:y1+h1, x1:x1+w1]
        eyes = eye_cascade.detectMultiScale(face_rect, minNeighbors=10)
        for eye in eyes:
            x2, y2, w2, h2 = eye
            center = int(x2+w2/2), int(y2+h2/2)
            cv2.circle(face_rect, center, int(w2/2), (255, 0, 0), 2)      
            
    cv2.imshow('camera', frame) # 재생    
       
    key = cv2.waitKey(delay) # delay(ms) 기다리기(sleep 효과)
    if key == 27: # 27 : Esc key,  종료조건    
        break

if cap.isOpened(): # True
    print('cap release!!')
    cap.release()
    
cv2.destroyAllWindows()

640.0 480.0 30.0 33
cap release!!

실습 : Blurring을 이용하여 얼굴만 가리기

import sys

face_cascade = cv2.CascadeClassifier('./data/haarcascade_frontalface_default.xml')

cap = cv2.VideoCapture(0) # 0: 메인 카메라

if not cap.isOpened():
    print("camera open failed")
    sys.exit()
    
width = cap.get(cv2.CAP_PROP_FRAME_WIDTH)    
height = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)    
fps = cap.get(cv2.CAP_PROP_FPS) # frame per second
delay = round(1000/30)

print(width, height, fps, delay)

while True:    
    ret, frame = cap.read() # frame : 이미지 한장 (shape : height x width x channel)        
    
    if not ret:
        print("frame read error")
        break
        
    # detect face and eyes
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray, minNeighbors=10)
    
    for face in faces:
        x1, y1, w1, h1 = face
        cv2.rectangle(frame, (x1, y1), (x1+w1, y1+h1), (0, 0, 255), 2)
        face_rect = frame[y1:y1+h1, x1:x1+w1]
        kernel_size = 50
        frame[y1:y1+h1, x1:x1+w1] = cv2.blur(face_rect, (kernel_size, kernel_size))          
            
    cv2.imshow('camera', frame) # 재생    
       
    key = cv2.waitKey(delay) # delay(ms) 기다리기(sleep 효과)
    if key == 27: # 27 : Esc key,  종료조건    
        break

if cap.isOpened(): # True
    print('cap release!!')
    cap.release()
    
cv2.destroyAllWindows()

640.0 480.0 30.0 33
cap release!!

참고 numpy stack

https://numpy.org/devdocs/reference/generated/numpy.dstack.html

np.dstack()

a = np.array([[1],[2],[3]])
b = np.array([[2],[3],[4]])
c = np.array([[3],[4],[5]])
print(a.shape, b.shape, c.shape)
d = np.dstack((a,b,c))
print(d.shape)

(3, 1) (3, 1) (3, 1)
(3, 1, 3)

np.hstack()

a = np.array([[1, 1],[2, 2],[3, 3]])
b = np.array([[4, 4],[5, 5],[6, 5]])
print(a.shape, b.shape)
c = np.hstack((a,b))
print(c.shape)

(3, 2) (3, 2)
(3, 4)

np.vstack()

a = np.array([[1], [2], [3]])
b = np.array([[4], [5], [6]])
print(a.shape, b.shape)
c = np.vstack((a,b))
print(c.shape)

(3, 1) (3, 1)
(6, 1)

Reference

OpenCV 4로 배우는 컴퓨터 비전과 머신 러닝 (황선규 저)

Twitter Facebook LinkedIn

Computer vision 7 (모폴로지 연산과 객체 검출)

모폴로지 연산과 객체 검출

1. 모폴로지 연산

침식과 팽창

열기와 닫기

2. 레이블링과 외곽선 검출

레이블링 기본

레이블링 응용

외곽선 검출과 그리기

실습 thumbs_up_down.jpg의 손 외곽선 그리기

3.객체 검출

캐스케이드 분류기와 얼굴 검출

실습 : multi face 검출

실습 : Camera Device로부터 얻은 frame에 대해 얼굴(눈) 검출하기

실습 : Blurring을 이용하여 얼굴만 가리기

Reference

공유하기

댓글남기기

참고

소개(about me)

Rnn 20 (attention and seq2seq learning using date dataset in pytorch)

Rnn 19 (attention and seq2seq learning using addition dataset in pytorch)

Rnn 18 (어텐션)