MachineLearning/final_data_processing.py at main · ttenneb/MachineLearning · 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
import cv2
import numpy as np
from math import sqrt
from matplotlib import pyplot as plt
import pickle

image_original = cv2.imread("Wood_Masked.jpg")
# fix the colors (BGR to RGB)
image_original = np.flip(image_original, axis=-1)

# get rid of the white part so we can find the colors
image = np.array_split(image_original, 3)
image = np.stack(image)
image = np.array_split(image, 3, axis=2)
image= np.stack(image)
image = list(image.reshape(9, 300, 300, 3))
del image[4]
image = np.concatenate(np.stack(image), axis=0)

# number of bins/colors to find
bin_count = 4
min_percent = .005

show_image = False
show_palette = True

try:
    bins = pickle.load(open("bins.p", "rb"))
except:
    bins = None
binned_pixels = []
total_pixels = len(image.reshape(-1, 3))*(8/9)
if bins is None or len(bins) != bin_count:
    print("Defining Bins")
    # must be square
    bins = np.random.randint(high=256, low = 0, size=(bin_count, 3))
    # find nearest bin for each value

    for iteration in range(100):
        binned_pixels = [[] for i in range(bin_count)]
        print("Finding Nearest Bin for Each Value")
        # find distances for each bin for each pixel
        dist = np.sum((bins[:, None, :]-image.reshape(-1, 3))**2, axis=2)
        # get closest bin for each pixel
        idxs = np.argmin(dist, axis=0)
        # assign pixel to bin
        for i, pixel in enumerate(image.reshape(-1, 3)):
            binned_pixels[idxs[i]].append(pixel)

        print("Calculating Mean for Each Bin")
        # calculate mean for each bin and redefine bin value
        total_zeros = 0
        max = 0
        max_bin = 0

        for i, bin in enumerate(binned_pixels):
            print(len(bin)/total_pixels)
            if len(bin)/total_pixels > min_percent:
                bins[i] = np.mean(bin, axis=0)
                if len(bin) > max:
                    max = len(bin)
                    max_bin = i
            else:
                bins[i] = np.random.randint(high=256, low = 0, size=(3))
                total_zeros += 1
        if total_zeros == 0:
            break
        # stop pixels from collecting at local maximum
        bins[max_bin] = np.random.randint(high=256, low = 0, size=(3))
        print("Iteration: ", iteration, " total_zeros: ", total_zeros)


if show_palette:
    plt.imshow(bins.reshape(1, -1, 3).astype(np.uint8))
    plt.show()

if show_image:
    # find nearest bin for each value
    dist = np.sum((bins[:, None, :]-image_original.reshape(-1, 3))**2, axis=2)
    idxs = np.argmin(dist, axis=0)

    # Map the image to the nearest colors
    quantized = bins[idxs]

    plt.imshow(quantized.reshape(image_original.shape).astype(np.uint8))
    plt.show()

pickle.dump(bins, open("bins.p", "wb"))

bin_sizes = {}


if len(binned_pixels) != 0:
    for i, bin in enumerate(binned_pixels):
        bin_sizes.update({i: len(bin)/total_pixels})

    # print(bin_sizes)
    pickle.dump(bin_sizes, open("bin_sizes.p", "wb"))

def quantize(image, bins = bins):
    # find nearest bin for each value
    dist = np.sum((bins[:, None, :]-image.reshape(-1, 3))**2, axis=2)
    idxs = np.argmin(dist, axis=0)

    # Map the image to the nearest colors
    quantized = bins[idxs]

    return quantized.reshape(image.shape).astype(np.uint8)

def quantize_indexs(image, bins = bins):
    # find nearest bin for each value
    dist = np.sum((bins[:, None, :]-image.reshape(-1, 3))**2, axis=2)
    idxs = np.argmin(dist, axis=0)

    # Map the image to the nearest colors
    quantized = idxs

    return quantized.reshape(image.shape[0], image.shape[1], 1).astype(np.uint8)

# print(bins)

def get_original_data():
    image_original = cv2.imread("Wood_Masked.jpg")
    # fix the colors (BGR to RGB)
    image_original = np.flip(image_original, axis=-1)

    # get rid of the white part so we can find the colors
    image = np.array_split(image_original, 3)
    image = np.stack(image)
    image = np.array_split(image, 3, axis=2)
    image= np.stack(image)
    image = list(image.reshape(9, 300, 300, 3))
    del image[4]

    processed_data = []
    for i, image in enumerate(image):
        processed_data.append(image)
        # for r in range(4):
        #     processed_data.append(np.rot90(image))

    image = np.concatenate(np.stack(processed_data), axis=0)
    return image