Upload 21 files

7501dfb verified 2 months ago

3.9 kB

from PIL import Image
import cv2
import numpy as np
def crop1(image_path, mask_path):
  # 读取图像和mask文件
  image = Image.open(image_path).convert("RGB")
  mask = Image.open(mask_path)

  # 确保图片和mask有相同的尺寸
  #assert image.size == mask.size

  # 将mask转换为二值图像
  mask = mask.convert('1')

  # 使用mask获取感兴趣的区域
  image = np.array(image)
  mask = np.array(mask)
  
  region_of_interest = image * mask[:, :, None]
  image=Image.fromarray(region_of_interest)
  return image




def crop2(image_path, mask_path):
    # 读取图像和mask
    image = Image.open(image_path).convert("RGBA")
    mask = Image.open(mask_path).convert('L')
    print(mask.size)

    # 确保图像和mask的大小相同
    assert image.size == mask.size, "图像和mask的大小必须相同"

    # 将图像和mask转换为numpy数组
    image_array = np.array(image)
    mask_array = np.array(mask)
    #mask_array[mask_array != 0] = 255

    #
    # print(sum(mask_array))


    # 创建一个新的透明度通道，对应mask为0的区域为透明
    alpha_channel = np.ones(mask_array.shape[:2], dtype=np.uint8) * 255
    print(mask_array.shape)
    alpha_channel[mask_array==0] = 0
    #mask2=Image.fromarray(alpha_channel)

   # print(mask_array)
   # print(alpha_channel)
    # 将新的透明度通道添加到图像中
    image_array[..., 3] = alpha_channel


    # 将结果数组转换回Image对象
    result_image = Image.fromarray(image_array)

    # 保存结果图像为PNG格式
    return result_image


def crop3(image_path, mask_path):
    # 读取图像和mask
    image = cv2.imread(image_path)
    if isinstance(mask_path, str):
        mask = cv2.imread(mask_path, cv2.IMREAD_GRAYSCALE)
    else:
        mask = np.ones((512, 512), dtype=np.uint8)

    # 计算mask的外接矩形
    # 计算mask的所有外接矩形
    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    min_x, min_y, max_x, max_y = None, None, None, None
    for contour in contours:
        x, y, w, h = cv2.boundingRect(contour)
        if min_x is None or x < min_x:
            min_x = x
        if min_y is None or y < min_y:
            min_y = y
        if max_x is None or x + w > max_x:
            max_x = x + w
        if max_y is None or y + h > max_y:
            max_y = y + h

    # 只保留原图对应的外接矩形的部分
    result = image[min_y:max_y, min_x:max_x]
    result=Image.fromarray(result)
    width, height = result.size
    if width < height:
        new_size = (224, int(height * 224 / width))
    else:
        new_size = (int(width * 224 / height), 224)

    # 将短边等比缩放到224
    result = result.resize(new_size, Image.Resampling.LANCZOS)
    return result
# 使用方法
#get_bounding_box('your_image.png', 'your_mask.png')
#
def crop4(image_path, mask_path):
    # 读取图像和mask
    image = cv2.imread(image_path)
    mask = cv2.imread(mask_path, cv2.IMREAD_GRAYSCALE)

    # 使用mask将对应图像mask=0的部分变黑
    image = image * np.stack([mask]*3, axis=-1)

    # 计算mask的外接矩形
    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    min_x, min_y, max_x, max_y = None, None, None, None
    for contour in contours:
        x, y, w, h = cv2.boundingRect(contour)
        if min_x is None or x < min_x:
            min_x = x
        if min_y is None or y < min_y:
            min_y = y
        if max_x is None or x + w > max_x:
            max_x = x + w
        if max_y is None or y + h > max_y:
            max_y = y + h

    # 只保留原图对应的外接矩形的部分
    result = image[min_y:max_y, min_x:max_x]
    result = Image.fromarray(result)

    return result

	from PIL import Image
	import cv2
	import numpy as np
	def crop1(image_path, mask_path):
	# 读取图像和mask文件
	image = Image.open(image_path).convert("RGB")
	mask = Image.open(mask_path)

	# 确保图片和mask有相同的尺寸
	#assert image.size == mask.size

	# 将mask转换为二值图像
	mask = mask.convert('1')

	# 使用mask获取感兴趣的区域
	image = np.array(image)
	mask = np.array(mask)

	region_of_interest = image * mask[:, :, None]
	image=Image.fromarray(region_of_interest)
	return image




	def crop2(image_path, mask_path):
	# 读取图像和mask
	image = Image.open(image_path).convert("RGBA")
	mask = Image.open(mask_path).convert('L')
	print(mask.size)

	# 确保图像和mask的大小相同
	assert image.size == mask.size, "图像和mask的大小必须相同"

	# 将图像和mask转换为numpy数组
	image_array = np.array(image)
	mask_array = np.array(mask)
	#mask_array[mask_array != 0] = 255

	#
	# print(sum(mask_array))


	# 创建一个新的透明度通道，对应mask为0的区域为透明
	alpha_channel = np.ones(mask_array.shape[:2], dtype=np.uint8) * 255
	print(mask_array.shape)
	alpha_channel[mask_array==0] = 0
	#mask2=Image.fromarray(alpha_channel)

	# print(mask_array)
	# print(alpha_channel)
	# 将新的透明度通道添加到图像中
	image_array[..., 3] = alpha_channel


	# 将结果数组转换回Image对象
	result_image = Image.fromarray(image_array)

	# 保存结果图像为PNG格式
	return result_image


	def crop3(image_path, mask_path):
	# 读取图像和mask
	image = cv2.imread(image_path)
	if isinstance(mask_path, str):
	mask = cv2.imread(mask_path, cv2.IMREAD_GRAYSCALE)
	else:
	mask = np.ones((512, 512), dtype=np.uint8)

	# 计算mask的外接矩形
	# 计算mask的所有外接矩形
	contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
	min_x, min_y, max_x, max_y = None, None, None, None
	for contour in contours:
	x, y, w, h = cv2.boundingRect(contour)
	if min_x is None or x < min_x:
	min_x = x
	if min_y is None or y < min_y:
	min_y = y
	if max_x is None or x + w > max_x:
	max_x = x + w
	if max_y is None or y + h > max_y:
	max_y = y + h

	# 只保留原图对应的外接矩形的部分
	result = image[min_y:max_y, min_x:max_x]
	result=Image.fromarray(result)
	width, height = result.size
	if width < height:
	new_size = (224, int(height * 224 / width))
	else:
	new_size = (int(width * 224 / height), 224)

	# 将短边等比缩放到224
	result = result.resize(new_size, Image.Resampling.LANCZOS)
	return result
	# 使用方法
	#get_bounding_box('your_image.png', 'your_mask.png')
	#
	def crop4(image_path, mask_path):
	# 读取图像和mask
	image = cv2.imread(image_path)
	mask = cv2.imread(mask_path, cv2.IMREAD_GRAYSCALE)

	# 使用mask将对应图像mask=0的部分变黑
	image = image * np.stack([mask]*3, axis=-1)

	# 计算mask的外接矩形
	contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
	min_x, min_y, max_x, max_y = None, None, None, None
	for contour in contours:
	x, y, w, h = cv2.boundingRect(contour)
	if min_x is None or x < min_x:
	min_x = x
	if min_y is None or y < min_y:
	min_y = y
	if max_x is None or x + w > max_x:
	max_x = x + w
	if max_y is None or y + h > max_y:
	max_y = y + h

	# 只保留原图对应的外接矩形的部分
	result = image[min_y:max_y, min_x:max_x]
	result = Image.fromarray(result)

	return result