def read_test_xy(filenames=None):
n = len(filenames)
batch_size = n
batch_x = np.zeros([batch_size, Config.image_h * Config.image_w])
batch_y = np.zeros([batch_size, Config.max_captcha * Config.char_set_len])
def get_image_sample(file):
# image = cv2.imread(file)
image = Image.open(file)
# image.show()
image = np.array(image)
filename = os.path.basename(file)
text = filename.replace('.png', '')
return text, image
for i in range(batch_size):
text, image = get_image_sample(filenames[i])
# 去噪、灰度、二值化处理
image = nd.noise_del(image)
image = cut_box(image, resize=(Config.image_w, Config.image_h)) # 剪切
image = nd.convert2gray(image) # 三维变2维
# image = nd.binarizing(image, threshold=200, cov=False)
# image1 = Image.fromarray(image)
# image1.show()
# batch_x[i, :] = (image.flatten() - 128) / 128 # image.flatten() / 255 # mean为0
batch_x[
i, :] = image.flatten() / 255 # (image.flatten()-128)/128 mean为0
batch_y[i, :] = text2vec(text)
return batch_x, batch_y
def get_next_batch(batch_size=128):
batch_x = np.zeros([batch_size, Config.image_h * Config.image_w])
batch_y = np.zeros([batch_size, Config.max_captcha * Config.char_set_len])
# 有时生成图像大小不是(60, 160, 3)
def wrap_gen_captcha_text_and_image():
while True:
text = random_captcha_text(captcha_size=4)
image = gen_captcha_text_and_image(text)
if image.shape == (30, 106, 3):
return text, image
for i in range(batch_size):
text, image = wrap_gen_captcha_text_and_image()
if batch_size == 1:
print('test text:', text)
# image1 = Image.fromarray(image)
# image1.show()
# 去噪、灰度、二值化处理
image = nd.noise_del(image)
image = cut_box(image,
resize=(Config.image_w, Config.image_h)) # 剪切最小框图
image = nd.convert2gray(image) # 三维变2维
# image = nd.binarizing(image,threshold=200, cov=False)
# image1 = Image.fromarray(image)
# image1.show()
batch_x[
i, :] = image.flatten() / 255 # (image.flatten()-128)/128 mean为0
batch_y[i, :] = text2vec(text)
return batch_x, batch_y
def predict_img(img, clf):
text = ''
image = nd.noise_del(img)
image = cut_box(image)
image_list = seg_img(image)
for im in image_list:
im = nd.convert2gray(im)
im = im.reshape(-1)
c = clf.predict([im,])[0]
text += c
return text
def img_cut_to_arry(img):
imgs = []
image = nd.noise_del(img)
# ima = Image.fromarray(image,'RGB')
# ima.show()
image = cut_box(image)
# ima = Image.fromarray(image)
# ima.show()
image_list = seg_img(image)
for im in image_list:
im = nd.convert2gray(im)
# ima = Image.fromarray(im)
# ima.show()
im = im.reshape(-1)
imgs.append(im)
return imgs
def get_text_xy(text):
batch_x = np.zeros([1, Config.image_h * Config.image_w])
batch_y = np.zeros([1, Config.max_captcha * Config.char_set_len])
image = gen_captcha_text_and_image(text)
# 去噪、灰度、二值化处理
image = nd.noise_del(image)
image = cut_box(image, resize=(Config.image_w, Config.image_h)) # 剪切
image = nd.convert2gray(image) # 三维变2维
# image = nd.binarizing(image, threshold=200, cov=False)
# image1 = Image.fromarray(image)
# image1.show()
batch_x[0, :] = image.flatten() / 255 # (image.flatten()-128)/128 mean为0
batch_y[0, :] = text2vec(text)
return batch_x, batch_y