def judge_vowel(image, index, vewol):
row, col = image.shape
new_vewol = []
if (index > 10):
image_vowel = image[0:index, :]
for i in range(col):
image[index][i] = 0
lb_image = measure.label(image_vowel, connectivity=1)
regions = measure.regionprops(lb_image)
k = 1
num = len(regions)
for region in regions:
if (region.area < 30):
continue
min_row, min_col, max_row, max_col = region.bbox
if (max_col - min_col) / (max_row - min_row) > 2.5:
for i in range(len(vewol)):
x, y = vewol[i][0], vewol[i][1]
print('x y yyyyyyyyyyyy', x, y)
print(min_row, min_col, max_row, max_col)
print("Using SVM to predict...")
if (max_row - min_row < 6):
break
images = segmentation_image(region.image, x - min_col,
y - min_row)
if (len(images) < 2):
break
image1 = images[0]
image2 = images[1]
print([x, y])
# plt.subplot(2, 2, 1).imshow(image1.astype(np.int8))
ratio1, classification = get_class_and_ratio(
image1.astype(np.uint8) * 255)
print(1, ratio1, classification)
# plt.subplot(2, 2, 2).imshow(image2.astype(np.int8))
ratio2, classification = get_class_and_ratio(
image2.astype(np.uint8) * 255)
print(2, ratio2, classification)
if (ratio1 > 0 and ratio2 > 0):
new_vewol.append([x, y])
# plt.subplot(212).imshow(image)
# plt.show()
# else:
# plt.subplot(111).imshow(image)
# plt.show()
# return False,image
return new_vewol
def judge_vowel(image, index):
row, col = image.shape
if (index > 10):
image_vowel = image[0:index - 2, :]
for i in range(col):
image[index - 2][i] = 0
lb_image = measure.label(image_vowel, connectivity=1)
regions = measure.regionprops(lb_image)
k = 1
num = len(regions)
for region in regions:
print("坎坎坷咳咳咳扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩扩")
plt.subplot(2, num, k).imshow(region.image.astype(np.int8))
ratio, classification = get_class_and_ratio(
region.image.astype(np.uint8) * 255)
print(k, ratio, classification)
k = k + 1
plt.subplot(212).imshow(image)
plt.show()
else:
plt.subplot(111).imshow(image)
plt.show()
return False, image