def test_canny_edge():
"""
canny边缘检测。
:return:
"""
image_file = os.path.join(image_dir, "demo1.png")
img = cv2.imread(image_file)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 高斯模糊,降低噪声
blurred = cv2.GaussianBlur(img, (3, 3), 0)
# 灰度图像
gray = cv2.cvtColor(blurred, cv2.COLOR_RGB2GRAY)
# 图像梯度
x_grad = cv2.Sobel(gray, cv2.CV_16SC1, 1, 0)
y_grad = cv2.Sobel(gray, cv2.CV_16SC1, 0, 1)
# 计算边缘
# 50和150参数必须符合1：3或者1：2
edge_output = cv2.Canny(x_grad, y_grad, 50, 150)
cv2.imshow("edge", edge_output)
dst = cv2.bitwise_and(img, img, mask=edge_output)
cv2.imshow('cedge', dst)
images = [img, blurred, gray, edge_output, dst]
pil_image_demo.plt_images(images)
def test_filter():
"""
模糊/平滑。
:return:
"""
lena_img = misc.imread(os.path.join(image_dir, "lena_gray.jpg"))
# print_image_pixel(lena_img)
# lena_img = Image.fromarray(lena_img)
# lena_img.show()
# 高斯滤镜
blurred_lena_none = ndimage.gaussian_filter(lena_img, sigma=0)
blurred_lena = ndimage.gaussian_filter(lena_img, sigma=3)
very_blurred = ndimage.gaussian_filter(lena_img, sigma=5)
# 均匀滤镜
local_mean = ndimage.uniform_filter(lena_img, size=11)
# 锐化处理，通过增加拉普拉斯近似增加边缘权重
blurred_l = ndimage.gaussian_filter(lena_img, 3)
filter_blurred_l = ndimage.gaussian_filter(blurred_l, 1)
alpha = 30
sharpened = blurred_l + alpha * (blurred_l - filter_blurred_l)
# 消噪处理
noisy = lena_img + 0.4 * lena_img.std() * np.random.random(lena_img.shape)
gauss_denoised = ndimage.gaussian_filter(noisy, 2)
med_denoised = ndimage.median_filter(noisy, 3)
# 显示图像
image_arrays = [
lena_img, blurred_lena_none, blurred_lena, very_blurred, local_mean,
sharpened, noisy, gauss_denoised, med_denoised
]
images = ndarray_2_image(image_arrays)
pil_image_demo.plt_images(images, 3)
def test_threshold():
"""
测试图像阈值。
:return:
"""
image_file1 = os.path.join(image_dir, "demo1.png")
image1 = cv2.imread(image_file1)
# 把输入图像灰度化
image1 = cv2.cvtColor(image1, cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(image1, cv2.COLOR_RGB2GRAY)
# gray = cv2.cvtColor(image1, cv2.COLOR_BGR2GRAY)
# show_image(gray)
# 直接阈值化是对输入的单通道矩阵逐像素进行阈值分割。
ret, binary1 = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_TRIANGLE)
# show_image(binary)
# 自适应阈值化能够根据图像不同区域亮度分布，改变阈值
binary2 = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 25, 10)
# show_image(binary)
# 根据平均像素值
h, w = gray.shape[:2]
m = np.reshape(gray, [1, w*h])
mean = m.sum()/(w*h)
common_logger.info("mean:{0}".format(mean))
ret, binary3 = cv2.threshold(gray, mean, 255, cv2.THRESH_BINARY)
# show_image(binary)
images = [image1, gray, binary1, binary2, binary3]
pil_image_demo.plt_images(images)
def plt_image_test():
"""
显示图像。
:return:
"""
im_nums = np.arange(8)
im = np.zeros([8, 8])
images = []
for im_num in im_nums:
im = im.copy()
im[im_num, im_num] = im_num
images.append(im)
pil_image_demo.plt_images(images, 3, interpolation='nearest')
def convert_image_color():
"""
图像颜色转换。
:return:
"""
image_file1 = os.path.join(image_dir, "demo1.png")
image1 = cv2.imread(image_file1)
image_cvt1 = cv2.cvtColor(image1, cv2.COLOR_BGR2HSV)
image_cvt2 = cv2.cvtColor(image1, cv2.COLOR_RGB2HSV)
image_cvt3 = cv2.cvtColor(image1, cv2.COLOR_BGR2RGB)
image_cvt4 = cv2.cvtColor(image1, cv2.COLOR_BGR2GRAY)
image_cvt5 = cv2.cvtColor(image_cvt3, cv2.COLOR_RGB2GRAY)
images = [image1, image_cvt1, image_cvt2, image_cvt3, image_cvt4, image_cvt5]
pil_image_demo.plt_images(images)
def gray_dilation():
"""
灰度图的形态修改
:return:
"""
# 灰度值图像
im = np.zeros((64, 64))
np.random.seed(2)
x, y = (63 * np.random.random((2, 8))).astype(np.int)
im[x, y] = np.arange(8)
# print_image_pixel(im)
# 灰度膨胀
bigger_points = ndimage.grey_dilation(im,
size=(5, 5),
structure=np.ones((5, 5)))
# print_image_pixel(bigger_points)
square = np.zeros((16, 16))
square[4:-4, 4:-4] = 1
dist = ndimage.distance_transform_bf(square)
dilate_dist = ndimage.grey_dilation(dist,
size=(3, 3),
structure=np.ones((3, 3)))
images = [im, bigger_points, square, dist, dilate_dist]
pil_image_demo.plt_images(images, 3)
plt.figure(figsize=(12.5, 3))
plt.subplot(141)
plt.imshow(im, interpolation='nearest')
plt.axis('off')
plt.subplot(142)
plt.imshow(bigger_points, interpolation='nearest')
plt.axis('off')
plt.subplot(143)
plt.imshow(dist, interpolation='nearest')
plt.axis('off')
plt.subplot(144)
plt.imshow(dilate_dist, interpolation='nearest')
plt.axis('off')
plt.subplots_adjust(wspace=0,
hspace=0.02,
top=0.99,
bottom=0.01,
left=0.01,
right=0.99)
plt.show()
def test_histogram():
"""
测试直方图。
:return:
"""
# 构造图像
n = 10
length = 256
np.random.seed(1)
im = np.zeros((length, length))
points = length * np.random.random((2, n**2))
im[(points[0]).astype(np.int), (points[1]).astype(np.int)] = 1
im = ndimage.gaussian_filter(im, sigma=length / (4. * n))
mask = (im > im.mean()).astype(np.float)
mask += 0.1 * im
img = mask + 0.2 * np.random.randn(*mask.shape)
print("max:{0},min:{1}".format(np.max(img), np.min(img)))
hist, bin_edges = np.histogram(img, bins=60)
print("hist:{0}".format(hist))
print("bin_edges:{0}".format(bin_edges))
bin_centers = 0.5 * (bin_edges[:-1] + bin_edges[1:])
binary_img = img > 0.5
images = [im, mask, binary_img]
pil_image_demo.plt_images(images, 3)
# 图像直方图
plt.figure(figsize=(11, 4))
plt.subplot(111)
plt.plot(bin_centers, hist, lw=2)
plt.axvline(0.5, color='r', ls='--', lw=2)
plt.text(0.57,
0.8,
'histogram',
fontsize=20,
transform=plt.gca().transAxes)
plt.yticks([])
plt.subplots_adjust(wspace=0.02,
hspace=0.3,
top=1,
bottom=0.1,
left=0,
right=1)
plt.show()
def edge_detect():
"""
边缘检测
:return:
"""
# 合成数据
im = np.zeros((256, 256))
im[64:-64, 64:-64] = 1
im = ndimage.rotate(im, 15, mode='constant')
im_gf = ndimage.gaussian_filter(im, 8)
# sobel检测
sx = ndimage.sobel(im_gf, axis=0, mode='constant')
sy = ndimage.sobel(im_gf, axis=1, mode='constant')
sob = np.hypot(sx, sy)
images = [im, im_gf, sx, sy, sob]
pil_image_demo.plt_images(images, 3)
def test_image_flip():
"""
翻转图像。
使用函数cv2.flip(img,flipcode)翻转图像，flipcode控制翻转效果。
flipcode = 0：沿x轴翻转
flipcode > 0：沿y轴翻转
flipcode < 0：x,y轴同时翻转
:return:
"""
image_file = os.path.join(image_dir, "demo1.png")
img = cv2.imread(image_file)
img_copy = img.copy()
img_flip1 = cv2.flip(img, 0)
img_flip2 = cv2.flip(img, 1)
img_flip3 = cv2.flip(img, -1)
images = [img_copy, img_flip1, img_flip2, img_flip3]
pil_image_demo.plt_images(images)
def test_measurement():
"""
测试测量对象属性。
:return:
"""
# 合成图像
n = 10
length = 256
im = np.zeros((length, length))
points = length * np.random.random((2, n**2))
im[(points[0]).astype(np.int), (points[1]).astype(np.int)] = 1
im = ndimage.gaussian_filter(im, sigma=length / (4. * n))
mask = im > im.mean()
# 标记连接成分
label_im, nb_labels = ndimage.label(mask)
print("区域数量：{0}".format(nb_labels))
# 计算每个区域的尺寸，均值等等
sizes = ndimage.sum(mask, label_im, range(nb_labels + 1))
mean_vals = ndimage.sum(im, label_im, range(1, nb_labels + 1))
print("各个区域的尺寸：{0}".format(sizes))
print("各个区域的平均尺寸：{0}".format(mean_vals))
# 计算尺寸小的连接成分
mask_size = sizes < 1000
remove_pixel = mask_size[label_im]
print(remove_pixel.shape)
label_im2 = label_im.copy()
label_im2[remove_pixel] = 0
# 使用np.searchsorted重新分配标签
labels = np.unique(label_im2)
label_im3 = np.searchsorted(labels, label_im2)
# 找到关注的封闭对象区域
slice_x, slice_y = ndimage.find_objects(label_im3 == 4)[0]
roi = im[slice_x, slice_y]
images = [im, mask, label_im, label_im2, label_im3, roi]
pil_image_demo.plt_images(images, 3)
def transform_shape():
"""
形状转换，剪裁，上下翻转，旋转。
:return:
"""
lena_img = imageio.imread(os.path.join(image_dir, "lena_gray.jpg"))
width, height = lena_img.shape
print("width:{0}, height:{1}".format(width, height))
# Cropping
crop_lena = lena_img[round(width / 4):round(-width / 4),
round(height / 4):round(-height / 4)]
# up <-> down flip
flip_ud_lena = np.flipud(lena_img)
# rotation
rotate_lena = ndimage.rotate(lena_img, 45)
rotate_lena_noreshape = ndimage.rotate(lena_img, 45, reshape=False)
images = [
lena_img, crop_lena, flip_ud_lena, rotate_lena, rotate_lena_noreshape
]
pil_image_demo.plt_images(images)
def test_image_trans():
"""
测试图像变换
:return:
"""
image_file = os.path.join(image_dir, "demo1.png")
img = cv2.imread(image_file)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
height, width, dim = img.shape
# 执行Gamma矫正，小于1的值让暗部细节大量提升，同时亮部细节少量提升
img_gamma = ImageEnhanceCV.gamma_transform(img, 0.5)
# 沿着横纵轴放大1.6倍，然后平移(-150,-240)，最后沿原图大小截取，等效于裁剪并放大
m_crop = np.array([
[1.6, 0, -150],
[0, 1.6, -240]
], dtype=np.float32)
img_crop = cv2.warpAffine(img, m_crop, (width, height))
# x轴的剪切变换，角度15°
theta = 15 * np.pi / 180
m_shear = np.array([
[1, np.tan(theta), 0],
[0, 1, 0]
], dtype=np.float32)
img_sheared = cv2.warpAffine(img, m_shear, (width, height))
# 顺时针旋转，角度15°
m_rotate = np.array([
[np.cos(theta), -np.sin(theta), 0],
[np.sin(theta), np.cos(theta), 0]
], dtype=np.float32)
img_rotated = cv2.warpAffine(img, m_rotate, (width, height))
# show images
images = [img, img_gamma, img_crop, img_sheared, img_rotated]
pil_image_demo.plt_images(images)
def test_morphology():
"""
测试图像的数学形态学。
:return:
"""
show_content = False
show_images = False
el = ndimage.generate_binary_structure(2, 1)
el_int = el.astype(np.int)
if show_content:
print(el)
print(el_int)
# 腐蚀：最小化滤镜
a = np.zeros((7, 7), dtype=np.int)
a[1:6, 2:5] = 1
a_erosion1 = ndimage.binary_erosion(a).astype(a.dtype)
a_erosion2 = ndimage.binary_erosion(a, structure=np.ones(
(5, 5))).astype(a.dtype)
if show_content:
print(a)
print(a_erosion1)
print(a_erosion2)
if show_images:
image_arrays = [a, a_erosion1, a_erosion2]
image_arrays = bin2gray(image_arrays)
images = ndarray_2_image(image_arrays)
pil_image_demo.plt_images(images, 3)
# 膨胀：最大化滤镜
a = np.zeros((5, 5))
a[2, 2] = 1
a_dilation = ndimage.binary_dilation(a).astype(a.dtype)
if show_content:
print(a)
print(a_dilation)
if show_images:
image_arrays = [a, a_dilation]
image_arrays = bin2gray(image_arrays)
images = ndarray_2_image(image_arrays)
pil_image_demo.plt_images(images, 3)
# 开操作：腐蚀 + 膨胀
square = np.zeros((32, 32))
square[10:-10, 10:-10] = 1
np.random.seed(2)
x, y = (32 * np.random.random((2, 20))).astype(np.int)
square[x, y] = 1
# 图像开操作
open_square = ndimage.binary_opening(square)
# 先腐蚀后膨胀
eroded_square = ndimage.binary_erosion(square)
reconstruction = ndimage.binary_propagation(eroded_square, mask=square)
if show_images:
image_arrays = [square, open_square, eroded_square, reconstruction]
image_arrays = bin2gray(image_arrays)
images = ndarray_2_image(image_arrays)
pil_image_demo.plt_images(images, 2)
