07/27 16:00

import numpy
import cv2
import pandas
import seaborn
import os
from matplotlib import pyplot

# constants
path = "./data/"
maxShrinkRate = 10


pictureList = []
for filename in os.listdir(path):
    if filename.endswith(".jpg"):
        pictureList.append(cv2.imread(path + filename, cv2.IMREAD_GRAYSCALE))


# 資料處理

def toGraph(originalGraphList, xshrinkRate, yshrinkRate):

    xSize = len(originalGraphList[0])
    ySize = len(originalGraphList[0][0])

    fixed_xSize = xSize // xshrinkRate
    fixed_ySize = ySize // yshrinkRate
    fixed_graph = numpy.zeros((fixed_xSize, fixed_ySize))

    for xi in range(fixed_xSize):
        for yi in range(fixed_ySize):
            for xj in range(xshrinkRate):
                for yj in range(yshrinkRate):
                    for graph in originalGraphList:
                        fixed_graph[xi][yi] += graph[xi * xshrinkRate + xj][yi * yshrinkRate + yj]
            fixed_graph[xi][yi] //= xshrinkRate * yshrinkRate * len(originalGraphList)

    fixed_result = numpy.array(pandas.value_counts(fixed_graph.flatten()))
    returnVar = numpy.arange(1, len(fixed_result) + 1)

    pyplot.figure(figsize=(15, 5))
    pyplot.subplot(1, 2, 1)
    pyplot.plot(returnVar, fixed_result)
    pyplot.title(f"{xshrinkRate} * {yshrinkRate}")
    pyplot.ylabel('appear time')
    pyplot.xlabel('rank')

    pyplot.subplot(1, 2, 2)
    pyplot.plot(returnVar, fixed_result)
    pyplot.xscale('log')
    pyplot.yscale('log')
    pyplot.title(f"{xshrinkRate} * {yshrinkRate}")
    pyplot.ylabel('appear time(log_10)')
    pyplot.xlabel('rank(log_10)')

    return [returnVar, fixed_result]


trans_pictureList = []

for i in pictureList:
    trans_pictureList.append(i.tolist())

fixed_resultList = []
for i in range(1, maxShrinkRate + 1):
    fixed_resultList.append(toGraph(trans_pictureList, i, i))  # here
#   fixed_resultList.append(toGraph(trans_pictureList, 1, i))
#   fixed_resultList.append(toGraph(trans_pictureList, i, 1))

# 畫總圖

pyplot.figure(figsize=(15, 5))
pyplot.subplot(1, 2, 1)
for i in fixed_resultList:
    pyplot.plot(i[0], i[1])
pyplot.title('total')
pyplot.ylabel('appear time')
pyplot.xlabel('rank')


pyplot.subplot(1, 2, 2)
for i in fixed_resultList:
    pyplot.plot(i[0], i[1])
pyplot.xscale('log')
pyplot.yscale('log')
pyplot.title('total')
pyplot.ylabel('appear time(log_10)')
pyplot.xlabel('rank(log_10)')


# 分析圖

logarithmic_resultList = []  # [x, y]
for i in fixed_resultList:
    logarithmic_resultList.append([numpy.log10(i[0]), numpy.log10(i[1])])


pyplot.figure(figsize=(15, 5))
pyplot.subplot(1, 2, 1)

for i in range(maxShrinkRate):
    u = seaborn.regplot(logarithmic_resultList[i][0],
                        logarithmic_resultList[i][1],
                        label=f"{i + 1} * {i + 1}")
u.set_title('analyze')
u.set_ylabel('appear time(log_10)')
u.set_xlabel('rank(log_10)')
u.legend(loc='upper right')

# 線條
pyplot.subplot(1, 2, 2)

for i in range(maxShrinkRate):
    v = seaborn.regplot(logarithmic_resultList[i][0],
                        logarithmic_resultList[i][1],
                        label=f"{i + 1} * {i + 1}",
                        scatter=False)
v.set_title('analyze')
v.set_ylabel('appear time(log_10)')
v.set_xlabel('rank(log_10)')
v.legend(loc='upper right')

# 取數值
pyplot.figure(figsize=(0.001, 0.001))


axList = []

for i in range(maxShrinkRate):
    pyplot.subplot(1, maxShrinkRate, i + 1)
    axList.append(seaborn.regplot(logarithmic_resultList[i][0],
                                  logarithmic_resultList[i][1],
                                  scatter=False))

slopeList = []

for i in range(maxShrinkRate):
    slope, intercept = numpy.polyfit(axList[i].get_lines()[0].get_xdata(),
                                     axList[i].get_lines()[0].get_ydata(),
                                     1)
    print(f"slope {i + 1} = {slope}")
    slopeList.append(slope)

print('the average of slopes =', sum(slopeList) / maxShrinkRate)