python实现图像随机裁剪的示例代码_Python

python实现图像随机裁剪的示例代码

2021-08-12 00:26我坚信阳光灿烂 Python

这篇文章主要介绍了python实现图像随机裁剪的示例代码，帮助大家更好的理解和使用python处理图片，感兴趣的朋友可以了解下

实验条件：

从1张图像随机裁剪100张图像
裁剪出图像的大小为 60 x 60
IoU 大于等于 th=0.6 的裁剪框用红色标出，其它裁剪框用蓝色标出
IoU 比对原始区域用绿框标出

实验代码：

				?

									import cv2 as cv 

									import numpy as np

									np.random.seed(0)

									# get IoU overlap ratio

									def iou(a, b):

									    # get area of a

									 area_a = (a[2] - a[0]) * (a[3] - a[1])

									    # get area of b

									 area_b = (b[2] - b[0]) * (b[3] - b[1])

									    # get left top x of IoU

									 iou_x1 = np.maximum(a[0], b[0])

									    # get left top y of IoU

									 iou_y1 = np.maximum(a[1], b[1])

									    # get right bottom of IoU

									 iou_x2 = np.minimum(a[2], b[2])

									    # get right bottom of IoU

									 iou_y2 = np.minimum(a[3], b[3])

									    # get width of IoU

									 iou_w = iou_x2 - iou_x1

									    # get height of IoU

									 iou_h = iou_y2 - iou_y1

									    # get area of IoU

									 area_iou = iou_w * iou_h

									    # get overlap ratio between IoU and all area

									 iou = area_iou / (area_a + area_b - area_iou)

									 return iou

									# crop and create database

									def crop_bbox(img, gt, Crop_N=200, L=60, th=0.5):

									 # get shape

									 H, W, C = img.shape

									 # each crop

									 for i in range(Crop_N):

									  # get left top x of crop bounding box

									  x1 = np.random.randint(W - L)

									  # get left top y of crop bounding box

									  y1 = np.random.randint(H - L)

									  # get right bottom x of crop bounding box

									  x2 = x1 + L

									  # get right bottom y of crop bounding box

									  y2 = y1 + L

									  # crop bounding box

									  crop = np.array((x1, y1, x2, y2))

									  # get IoU between crop box and gt

									  _iou = iou(gt, crop)

									  # assign label

									  if _iou >= th:

									   cv.rectangle(img, (x1, y1), (x2, y2), (0,0,255), 1)

									   label = 1

									  else:

									   cv.rectangle(img, (x1, y1), (x2, y2), (255,0,0), 1)

									   label = 0

									 return img

									# read image

									img = cv.imread("../xiyi.jpg")

									img1 = img.copy()

									# gt bounding box

									gt = np.array((87, 51, 169, 113), dtype=np.float32)

									# get crop bounding box

									img = crop_bbox(img, gt, Crop_N=100, L=60, th=0.6)

									# draw gt

									cv.rectangle(img, (gt[0], gt[1]), (gt[2], gt[3]), (0,255,0), 1)

									cv.rectangle(img1,(gt[0], gt[1]), (gt[2], gt[3]), (0,255,0), 1)

									cv.imshow("result1",img1)

									cv.imshow("result", img)

									cv.imwrite("out.jpg", img)

									cv.waitKey(0)

									cv.destroyAllWindows()