def avg():
file_name = gauss_noise_filename
header, rgb = bmp.readBmp(file_name)
psnr = []
rng = range(1, 25)
y_cb_cr = bmp.convertYCbCr(rgb)
orinal_y_cb_cr = bmp.convertYCbCr(orig_pic)
for i in rng:
res = BmpFilters.BmpFilters.avg(y_cb_cr, i)
psnr.append(bmp.psnr(orinal_y_cb_cr[..., 0], res[..., 0]))
res[..., 1] = res[..., 0]
res[..., 2] = res[..., 0]
# bmp.writeBmp(moving_average_folder + f"/avg {i}.bmp", header, res)
x = np.array(list(rng), dtype='uint')
y = np.array(psnr)
np.savetxt(moving_average_folder + "/psnr.csv", (x, y),
delimiter=',',
fmt='%.2f')
plt.plot(list(rng), psnr)
plt.savefig(moving_average_folder + "/psnr.png")
def laplas():
header, rgb = bmp.readBmp(orig_filename)
y_cb_cr = bmp.convertYCbCr(rgb)
y_cb_cr_res = BmpFilters.BmpFilters.laplas_operator(y_cb_cr, 0)
y_cb_cr_res += 128
np.putmask(y_cb_cr_res, y_cb_cr_res < 0, 0)
np.putmask(y_cb_cr_res, y_cb_cr_res > 255, 255)
y_cb_cr_res[..., 1] = y_cb_cr_res[..., 0]
y_cb_cr_res[..., 2] = y_cb_cr_res[..., 0]
y_cb_cr[..., 1] = y_cb_cr[..., 0]
y_cb_cr[..., 2] = y_cb_cr[..., 0]
bmp.writeBmp(laplas_folder + "/laplas.bmp", header, y_cb_cr_res)
bmp.writeBmp(laplas_folder + "/orig.bmp", header, y_cb_cr)
def impulse_noise():
# hist = defaultdict(list)
# x = defaultdict(list)
# hist[0.1].append(1)
# hist[0.1].append(2)
# x[0.1].append(3)
# x[0.1].append(6)
# x[0.1].append(6)
# for (k, v), (k2, v2) in zip(hist.items(), x.items()):
# print(k, v, k2, v2)
# print('\n')
#
# return
small = 0.00001
file_name = orig_filename
header, rgb = bmp.readBmp(file_name)
y_cb_cr = bmp.convertYCbCr(rgb)
hist = defaultdict(list)
x = defaultdict(list)
rng = np.arange(0, 1.1, 0.1)
for i in rng:
for j in rng:
if i + j >= 1:
continue
if i == j == 0:
i = small
j = small
res = BmpFilters.BmpFilters.add_impulse_noise(y_cb_cr, i, j)
p = bmp.psnr(y_cb_cr[..., 0], res[..., 0])
if i == j == small:
i = 0
j = 0
x[i].append(j)
hist[i].append(p)
res[..., 1] = res[..., 0]
res[..., 2] = res[..., 0]
bmp.writeBmp(
'impulse/impulse pa = {0:.1f} pb = {1:.1f}.png'.format(
float(i), float(j)), header, res)
l = []
for (k, v), (k1, v1) in zip(x.items(), hist.items()):
leg, = plt.plot(list(v),
list(v1),
label='pa = {0:.1}'.format(float(k)))
l.append(leg)
plt.legend(handles=l)
plt.savefig('impulse/psnr.png')
def gauss_noise():
header, rgb = bmp.readBmp(orig_filename)
y_cb_cr = bmp.convertYCbCr(rgb)
psnr = []
rng = range(0, 250, 10)
for i in rng:
if i == 0:
i = 0.0000001
res = BmpFilters.BmpFilters.add_gauss_noise(y_cb_cr, d=i, m=0)
res[..., 1] = res[..., 0]
res[..., 2] = res[..., 0]
p = bmp.psnr(y_cb_cr[..., 0], res[..., 0])
psnr.append(p)
if i == 0.0000001:
i = 0
bmp.writeBmp(f"gauss/gauss d ={i}.bmp", header, res)
plt.plot(list(rng), psnr)
plt.savefig(gauss_noise_psnr_filename)
def avgDecimationYCbCr(rgb, n):
print(f'\n \n \n avg decimation {n}')
y = rgb.shape[0]
x = rgb.shape[1]
y_cb_cr = bmp.convertYCbCr(rgb)
resG = bmp.decimationByAverageValue(y_cb_cr[..., 1], n)
resR = bmp.decimationByAverageValue(y_cb_cr[..., 2], n)
resG2 = bmp.recoverDecimationByDeletingEven(resG, n)
resR2 = bmp.recoverDecimationByDeletingEven(resR, n)
rgb_recovered = np.zeros((y, x, 3), 'uint8')
rgb_recovered[..., 0] = y_cb_cr[..., 0]
rgb_recovered[..., 1] = resG2
rgb_recovered[..., 2] = resR2
print("psnr cb", bmp.psnr(y_cb_cr[..., 1], resG2))
print("psnr cr", bmp.psnr(y_cb_cr[..., 2], resR2))
rgb_recovered = bmp.inverseConvertYCbCr(rgb_recovered)
print("psnr blue", bmp.psnr(rgb[..., 0], rgb_recovered[..., 0]))
print("psnr green", bmp.psnr(rgb[..., 1], rgb_recovered[..., 1]))
print("psnr red", bmp.psnr(rgb[..., 2], rgb_recovered[..., 2]))
return rgb_recovered
def componentHistogramsAndEntropy(rgb):
y_cb_cr = bmp.convertYCbCr(rgb)
x = range(0, 255)
freq_blue = defaultdict(int)
freq_green = defaultdict(int)
freq_red = defaultdict(int)
freq_cb = defaultdict(int)
freq_cr = defaultdict(int)
freq_y = defaultdict(int)
for b, g, r, cb, cr, y in zip(rgb[..., 0].flatten(), rgb[..., 1].flatten(),
rgb[..., 2].flatten(), y_cb_cr[...,
1].flatten(),
y_cb_cr[...,
2].flatten(), y_cb_cr[...,
0].flatten()):
freq_blue[b] += 1
freq_green[g] += 1
freq_red[r] += 1
freq_cb[cb] += 1
freq_cr[cr] += 1
freq_y[y] += 1
b = []
g = []
r = []
cb = []
cr = []
y = []
for i in x:
b.append(freq_blue[i])
g.append(freq_green[i])
r.append(freq_red[i])
cb.append(freq_cb[i])
cr.append(freq_cr[i])
y.append(freq_y[i])
size = rgb[..., 0].size
plt.bar(x, y)
plt.title("y")
plt.savefig("images\\y hist.png")
plt.clf()
plt.bar(x, b)
plt.title("blue")
plt.savefig("images\\blue hist.png")
plt.clf()
plt.bar(x, g)
plt.title("green")
plt.savefig("images\\green hist.png")
plt.clf()
plt.bar(x, r)
plt.title("red")
plt.savefig("images\\red hist.png")
plt.clf()
plt.bar(x, cb)
plt.title("cb")
plt.savefig("images\\cb hist.png")
plt.clf()
plt.bar(x, cr)
plt.title("cr")
plt.savefig("images\\cr hist.png")
plt.savefig("images\\cr hist.png")
plt.clf()
entropy_b = -np.sum(b * np.ma.log2(b).filled(0))
entropy_g = -np.sum(g * np.ma.log2(g).filled(0))
entropy_r = -np.sum(r * np.ma.log2(r).filled(0))
entropy_cb = -np.sum(cb * np.ma.log2(cb).filled(0))
entropy_cr = -np.sum(cr * np.ma.log2(cr).filled(0))
entropy_y = -np.sum(y * np.ma.log2(y).filled(0))
print("\n \n \n")
print("entropy b ", entropy_b)
print("entropy g ", entropy_g)
print("entropy r ", entropy_r)
print("entropy cb ", entropy_cb)
print("entropy cr ", entropy_cr)
print("entropy y ", entropy_y)
def DPCM(rgb, func_type):
y_cb_cr = bmp.convertYCbCr(rgb)
r = rgb[..., 2]
g = rgb[..., 1]
b = rgb[..., 0]
y = y_cb_cr[..., 0]
cb = y_cb_cr[..., 1]
cr = y_cb_cr[..., 2]
hist_d_red = r[1:, 1:].astype(np.int32) - bmp.DPCM(r, func_type)
hist_d_green = g[1:, 1:].astype(np.int32) - bmp.DPCM(g, func_type)
hist_d_blue = r[1:, 1:].astype(np.int32) - bmp.DPCM(b, func_type)
hist_d_y = r[1:, 1:].astype(np.int32) - bmp.DPCM(y, func_type)
hist_d_cb = r[1:, 1:].astype(np.int32) - bmp.DPCM(cb, func_type)
hist_d_cr = r[1:, 1:].astype(np.int32) - bmp.DPCM(cr, func_type)
freq_blue = defaultdict(int)
freq_green = defaultdict(int)
freq_red = defaultdict(int)
freq_cb = defaultdict(int)
size = rgb[..., 0].size
freq_cr = defaultdict(int)
freq_y = defaultdict(int)
for b, g, r, cb, cr, y in zip(hist_d_blue.flatten(),
hist_d_green.flatten(), hist_d_red.flatten(),
hist_d_cb.flatten(), hist_d_cr.flatten(),
hist_d_y.flatten()):
freq_blue[b] += 1
freq_green[g] += 1
freq_red[r] += 1
freq_cb[cb] += 1
freq_cr[cr] += 1
freq_y[y] += 1
b = []
g = []
r = []
cb = []
cr = []
y = []
x = range(-255, 255)
for i in x:
b.append(freq_blue[i])
g.append(freq_green[i])
r.append(freq_red[i])
cb.append(freq_cb[i])
cr.append(freq_cr[i])
y.append(freq_y[i])
for i in range(-255, 255):
b[i] = b[i] / size
g[i] = g[i] / size
r[i] = r[i] / size
cb[i] = cb[i] / size
cr[i] = cr[i] / size
y[i] = y[i] / size
plt.bar(x, b)
plt.title(f"BLUE DPCM type {func_type}")
plt.savefig(f"images\\BLUE DPCM type {func_type}.png")
plt.clf()
plt.bar(x, g)
plt.title(f"GREEN DPCM type {func_type}")
plt.savefig(f"images\\GREEN DPCM type {func_type}.png")
plt.clf()
plt.bar(x, r)
plt.title(f"RED DPCM type {func_type}")
plt.savefig(f"images\\RED DPCM type {func_type}.png")
plt.clf()
plt.bar(x, cb)
plt.title(f"CB DPCM type {func_type}")
plt.savefig(f"images\\CB DPCM type {func_type}.png")
plt.clf()
plt.bar(x, cr)
plt.title(f"CR DPCM type {func_type}")
plt.savefig(f"images\\CR DPCM type {func_type}.png")
plt.clf()
entropy_b = -np.sum(b * np.ma.log2(b).filled(0))
entropy_g = -np.sum(g * np.ma.log2(g).filled(0))
entropy_r = -np.sum(r * np.ma.log2(r).filled(0))
entropy_cb = -np.sum(cb * np.ma.log2(cb).filled(0))
entropy_cr = -np.sum(cr * np.ma.log2(cr).filled(0))
entropy_y = -np.sum(y * np.ma.log2(y).filled(0))
print("\n \n \n")
print(f"entropy type {func_type} of b ", entropy_b)
print(f"entropy type {func_type} of g ", entropy_g)
print(f"entropy type {func_type} of r ", entropy_r)
print(f"entropy type {func_type} of cb ", entropy_cb)
print(f"entropy type {func_type} of cr ", entropy_cr)
print(f"entropy type {func_type} of y ", entropy_y)
entropy_y = -np.sum(y * np.ma.log2(y).filled(0))
print("\n \n \n")
print(f"entropy type {func_type} of b ", entropy_b)
print(f"entropy type {func_type} of g ", entropy_g)
print(f"entropy type {func_type} of r ", entropy_r)
print(f"entropy type {func_type} of cb ", entropy_cb)
print(f"entropy type {func_type} of cr ", entropy_cr)
print(f"entropy type {func_type} of y ", entropy_y)
header, rgb = bmp.readBmp(in_file)
a = rgb[...][...][0] - rgb[...][...][1]
print(np.max(a))
sys.exit(0)
y_cb_cr = bmp.convertYCbCr(rgb)
cbDecimated = bmp.decimationByDeletingEven(y_cb_cr[..., 1], 4)
crDecimated = bmp.decimationByDeletingEven(y_cb_cr[..., 2], 4)
cbRecovered = bmp.recoverDecimationByDeletingEven(cbDecimated, 4)
crRecovered = bmp.recoverDecimationByDeletingEven(crDecimated, 4)
y_cb_cr[..., 1] = cbRecovered
y_cb_cr[..., 2] = crRecovered
rgb2 = bmp.inverseConvertYCbCr(y_cb_cr)
bmp.writeBmp(res_file_path + "testRec.bmp", header, rgb2)
print(bmp.psnr(rgb[..., 0], rgb2[..., 0]))
print(bmp.psnr(rgb[..., 1], rgb2[..., 1]))
print(bmp.psnr(rgb[..., 2], rgb2[..., 2]))
sys.stdout = open('console.txt', 'w')
b = rgb[..., 0]
g = rgb[..., 1]