Constellation-classifier/5_demo.py at main · PyoSH/Constellation-classifier · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Thu May 26 17:13:38 2022

@author: 표승현

검출기능 구현을 가정한 목표설명용 데모 코드
GET_ROI함수 상단에서 ROI를 바꾸고 main함수 상단의 im경로를 바꿔야 정상적으로 돌아간다.

"""
import cv2
import numpy as np
from tensorflow.keras.models import load_model  # 모델 불러오기 라이브러리

def GET_ROI(im):
    #################1. 입력 이미지에 맞게 ROI 바꾸기!!!!###############
    # (x,y),(w,h) =(92,382),(194,190)     ## ROI -> DemoImg_crux.jpeg
    # (x,y),(w,h) =(192,582),(194,190)     ## ROI -> nothing
    # (x,y),(w,h) =(20,238),(658,412)  ## ROI -> DemoImg_BigDipper.jpeg
    (x,y),(w,h) =(530,200),(400,430)  ## ROI -> Demo_BigDipper_2.jpeg
    roi_img = im[y:y+h, x:x+w]

    cv2.rectangle(im, (x,y,w,h), 255, 1)
    return (x,y,w,h)

def LINEDETECT(img):
    img_color = cv2.cvtColor(img,cv2.COLOR_GRAY2BGR)
    minLineLength = 100
    maxLineGap = 0

    lines = cv2.HoughLinesP(img,1,np.pi/360,133,minLineLength,maxLineGap)
    for i in range(len(lines)):
        for x1,y1,x2,y2 in lines[i]:
            cv2.line(img_color,(x1,y1),(x2,y2),(0,0,255),15)

    return img_color

def binarize_grpAll(gray_img):
    width = 320             # 320
    height = 240
    img_origin = gray_img.copy()

    ## 2. 임계값을 기준으로 이진화하기
    ret, th1 = cv2.threshold(gray_img,150,255, cv2.THRESH_BINARY)
    th3 = cv2.adaptiveThreshold(gray_img,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY_INV,15,15)
    t, t_otsu = cv2.threshold(gray_img, -1, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
    print('otsu threshold:', t)

    kernel = np.ones((13, 13), np.uint8)
    th3_close = cv2.morphologyEx(th3, cv2.MORPH_CLOSE, kernel)


    ## 2.1 FAST전처리
    fast = cv2.FastFeatureDetector_create()

    kp_2 = fast.detect(th3_close,None)
    FAST_2 =  cv2.cvtColor(cv2.drawKeypoints(th3_close, kp_2, None),cv2.COLOR_BGR2GRAY)

    kp_3 = fast.detect(t_otsu,None)
    FAST_3 =cv2.cvtColor(cv2.drawKeypoints(t_otsu, kp_3, None),cv2.COLOR_BGR2GRAY)
    Hough_0 = cv2.cvtColor(LINEDETECT(th3_close),cv2.COLOR_BGR2GRAY)

    dst_0 = cv2.resize(img_origin, dsize=(width, height), interpolation=cv2.INTER_LINEAR)
    dst_1 = cv2.resize(th1, dsize=(width, height), interpolation=cv2.INTER_LINEAR)
    dst_2 = cv2.resize(th3_close, dsize=(width, height), interpolation=cv2.INTER_LINEAR)
    dst_3 = cv2.resize(t_otsu, dsize=(width, height), interpolation=cv2.INTER_LINEAR)
    dst_4 = cv2.resize(FAST_2, dsize=(320, 240), interpolation=cv2.INTER_LINEAR)
    dst_5 = cv2.resize(FAST_3, dsize=(width, height), interpolation=cv2.INTER_LINEAR)
    dst_6 = cv2.resize(Hough_0, dsize=(width, height), interpolation=cv2.INTER_LINEAR)

    img_result = np.stack((dst_0,dst_1,dst_2,dst_3,dst_4,dst_5,dst_6),axis=-1)
    print("detect_shape= {}".format(img_result.shape))

    return img_result


if __name__ == '__main__':

    #################0. 입력 이미지 바꾸기!!!!###############
    # im = cv2.imread("DemoImg_BigDipper.jpeg",cv2.IMREAD_GRAYSCALE)
    im = cv2.imread("Demo_BigDipper_2.jpeg",cv2.IMREAD_GRAYSCALE)
    # im = cv2.imread("DemoImg_crux.jpeg",cv2.IMREAD_GRAYSCALE)

# ================테스트용 npy파일 만들기================================================
    (x,y,w,h) = GET_ROI(im) # ROI = 578,240 ~ 578,618 ~ 940,618 ~940,240
    roi_img = im[y:y+h, x:x+w]
#
    binarize_grpAll(im)
    img_detected = binarize_grpAll(roi_img)
    # np.save(("./demo_array.npy"), img_detected)
#
# ======================================================================================
    model = load_model('my_cnn_model_2018120133.h5') # 모델을 새로 불러옴

    img_input = img_detected
    # img_input = np.load("demo_array.npy")
    x_input = img_input.reshape(1, 240, 320, 7)

    prediction = model.predict(x_input) # 예측 수행

    constellation_class = {
      0: 'Big Dipper'
    , 1: 'Crux'
    , 2: 'Nothing'}

    class_predicted = np.argmax(prediction) # 클래스(예측)

    #################2. 입력 이미지에 맞게 정답 클래스 바꾸기!!!!###############
    class_actual = 0          # 클래스(정답) - 수동설정

    class_predicted_name = constellation_class[class_predicted] # 클래스 이름(예측)
    class_actual_name = constellation_class[class_actual] # 클래스 이름(정답)


    print("클래스(예측) : %s" %class_predicted_name) # 클래스 이름(예측) 출력
    print("클래스(정답) : %s" %class_actual_name) # 클래스 이름(정답) 출력

    import matplotlib.pyplot as plt
    # print(x_test[index,:,:,:])

    test_img = x_input[0,:,:,0].astype(np.uint8) # predict -> use for classification
    #################3. 입력 이미지에 맞게 Wide scene 이미지 바꾸기!!!!###############
    # img_wide =cv2.imread("DemoImg_crux.jpeg")
    # img_wide =cv2.imread("DemoImg_BigDipper.jpeg") #입력영상 바꾸면 바꿔줘야함!
    img_wide =cv2.imread("Demo_BigDipper_2.jpeg") #입력영상 바꾸면 바꿔줘야함!

    titles = ['Wide scene','ROI']
    images = [img_wide, test_img]

    for i in range(2):
            plt.subplot(1,2,i+1);plt.imshow(images[i])
            plt.title(titles[i])
            plt.xticks([]),plt.yticks([])
    plt.savefig("detect_table.png",dpi=400)