94 lines
2.0 KiB
Python
94 lines
2.0 KiB
Python
import math
|
|
import cv2 as cv
|
|
import numpy as np
|
|
from numba import jit
|
|
|
|
# camera settings
|
|
file = '01-0003.png'
|
|
I_Darkcurrent = 150.5
|
|
exposure_time = 0.5
|
|
f_stop = 2.4
|
|
ISO = 64 # basically brightness
|
|
|
|
# runtime config
|
|
MAX_GR_RATIO = 2000
|
|
MIN_GR_RATIO = None
|
|
|
|
x1 = 420
|
|
x2 = 1200
|
|
y1 = 400
|
|
y2 = -1
|
|
|
|
@jit(nopython=True)
|
|
def rg_ratio_normalize(imgarr):
|
|
imgnew = imgarr
|
|
for i in range(len(imgarr)):
|
|
for j in range(len(imgarr[i])):
|
|
px = imgarr[i][j]
|
|
r_norm = normalization_func(px[0])
|
|
g_norm = normalization_func(px[1])
|
|
|
|
# apply camera calibration func
|
|
ratio = pyrometry_calibration_formula(g_norm, r_norm)
|
|
|
|
# remove edge cases
|
|
if MAX_GR_RATIO != None and ratio > MAX_GR_RATIO or MIN_GR_RATIO != None and ratio < MIN_GR_RATIO:
|
|
ratio = 0
|
|
|
|
imgnew[i][j] = [ratio, ratio, ratio]
|
|
return imgnew
|
|
|
|
@jit(nopython=True)
|
|
def normalization_func(i):
|
|
return (i - I_Darkcurrent) * (f_stop ** 2) / (ISO * exposure_time)
|
|
|
|
@jit(nopython=True)
|
|
def pyrometry_calibration_formula(i_ng, i_nr):
|
|
return 362.73 * math.log10(
|
|
(i_ng/i_nr) ** 3
|
|
) + 2186.7 * math.log10(
|
|
(i_ng/i_nr) ** 3
|
|
) + 4466.5 * math.log10(
|
|
(i_ng / i_nr) ** 3
|
|
) + 3753.5
|
|
|
|
# read image & crop
|
|
file_name = file.split(".")[0]
|
|
file_ext = file.split(".")[1]
|
|
img = cv.imread(file)
|
|
img = img[y1:y2, x1:x2]
|
|
cv.imwrite(f'{file_name}-cropped.{file_ext}', img)
|
|
|
|
# img = cv.imread('ember_test.png')
|
|
|
|
img = rg_ratio_normalize(img)
|
|
|
|
# apply smoothing conv kernel
|
|
kernel = np.array([
|
|
[1/2, 1/2],
|
|
[1/2, 1/2],
|
|
])
|
|
|
|
# kernel = np.array([
|
|
# [1/3, 1/3, 1/3],
|
|
# [1/3, 1/3, 1/3],
|
|
# [1/3, 1/3, 1/3],
|
|
# ])
|
|
|
|
# kernel = np.array([
|
|
# [1/4, 1/4, 1/4, 1/4],
|
|
# [1/4, 1/4, 1/4, 1/4],
|
|
# [1/4, 1/4, 1/4, 1/4],
|
|
# [1/4, 1/4, 1/4, 1/4],
|
|
# ])
|
|
|
|
# Scaling adjustment factor
|
|
kernel *= 3/5
|
|
|
|
img = cv.filter2D(src=img, ddepth=-1, kernel=kernel)
|
|
|
|
# apply jet color map
|
|
img = cv.applyColorMap(img, cv.COLORMAP_JET)
|
|
|
|
cv.imwrite(f'{file_name}-cropped-transformed-ratio.{file_ext}', img)
|