refractometer/process_images.py at master · anfractuosity/refractometer · 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
137
138
139
140
141
142
143
144
#!/usr/bin/python3

import os
import cv2
import csv
import numpy as np
import math
from decimal import Decimal
from scipy import signal
import matplotlib.pyplot as plt

# Font for temperature text
font                   = cv2.FONT_HERSHEY_SIMPLEX
bottomLeftCornerOfText = (100,100)
fontScale              = 2
fontColor              = (0,0,255)
lineType               = 2

# Parameters used for cropping image
x = 1300
y = 50
w = 1024
h = 1500

# Top of scale
tpos = 182

# Zero position
zpos = 1350

# Process csv file and images of refractometer
#
# Outputs tuple of temperatures and distance of brix line from top of scale
def process_dataset(csvname, output):

    imcount = 0
    counter = 0

    temps = []
    diffs = []

    f = open(output, 'w')
    csv_dict = {"diff": -1, "temp": -1, "scaleheight": -1}
    wr = csv.DictWriter(f, csv_dict.keys())
    wr.writeheader()

    try:
        os.mkdir("/tmp/imgs")
    except FileExistsError:
        print("Delete /tmp/imgs first")
        quit()

    datasetfile = open(csvname)
    datasetreader = csv.reader(datasetfile)

    for row in datasetreader:

        tmp = Decimal(row[1]).quantize(Decimal("0.01"))
        textv = "%s C" % tmp

        filen = "dataset2/%s.jpg" % row[0]
        img = cv2.imread(filen, cv2.IMREAD_GRAYSCALE)
        if img is None:
            break

        rows,cols = img.shape
        M = cv2.getRotationMatrix2D((cols/2,rows/2),160,1)
        dst = cv2.warpAffine(img,M,(cols,rows))

        crop_img = dst[y:y+h, x:x+w]
        img = crop_img

        # Get mean across horizontal axis
        m = np.mean(crop_img[0:0+h, 0:0+w], axis=1)

        # Find last important peak
        sig, _ = signal.find_peaks(-m)
        prominences = signal.peak_prominences(-m, sig)[0]
        z = np.argwhere(prominences > 0.5)
        brix_line = sig[z[-1][0]]

        # Add lines denoting current position of brix line etc.
        crop_img = np.swapaxes(np.array([m]*w),0,1)
        crop_img = cv2.cvtColor(crop_img.astype(np.uint8), cv2.COLOR_GRAY2BGR)
        crop_img = cv2.line(crop_img, (0, brix_line), (w, brix_line), (0, 0, 255), 4)
        crop_img = cv2.line(crop_img, (0, zpos), (w, zpos), (0, 255, 0), 4)
        crop_img = cv2.line(crop_img, (0, tpos), (w, tpos), (255, 0, 0), 4)

        cv2.putText(crop_img,textv,
            bottomLeftCornerOfText,
            font,
            fontScale,
            fontColor,
            lineType)

        img = cv2.cvtColor(img.astype(np.uint8), cv2.COLOR_GRAY2BGR)
        img = cv2.line(img, (0, zpos), (w, zpos), (0, 255, 0), 4)
        img = cv2.line(img, (0, tpos), (w, tpos), (255, 0, 0), 4)

        # Only use first 100 images in dataset and only write image if detected line is below zero line (as it
        # must be during calibration)
        if counter <= 100 and brix_line > zpos:
            current_diff_from_top_scale = brix_line - tpos
            scale_height = zpos - tpos
            if True:
            #if len(diffs) == 0 or current_diff_from_top_scale > diffs[-1]:
                temps.append(float(tmp))
                diffs.append(current_diff_from_top_scale)
                csv_dict = {"diff": current_diff_from_top_scale, "temp": tmp, "scaleheight": scale_height}
                print(csv_dict)
                wr.writerow(csv_dict)
                f.flush()

                cv2.imwrite('/tmp/imgs/%s_a.jpg' % row[0], cv2.hconcat([crop_img, img]))
                cv2.imwrite('/tmp/imgs/%d.jpg' % imcount, cv2.hconcat([crop_img, img]))

                imcount += 1

        counter += 1

    return temps, diffs

def temp_correction(temps, diffs, temp):

    p = np.polyfit(temps, diffs, 3)
    polynomial = np.poly1d(p)

    x_axis = np.linspace(20,70)
    y_axis = polynomial(x_axis)
    plt.figure(1)
    plt.plot(x_axis, y_axis)
    plt.show()

    return np.poly1d(p)(temp)

if __name__ == "__main__":
    temps, diffs = process_dataset("dataset2.csv", "out.csv")

    current_diff_from_top_scale = 1284
    current_temp = 70.31
    scale_height = 1170

    brix = (1 - ((current_diff_from_top_scale - (temp_correction(temps, diffs, current_temp) - scale_height)) / scale_height)) * 32
    print(brix,"brix")