from PIL import Imagefrom itertools import groupbyimport cv2import datetimeimport os # from core.rabbitmq import MessageQueue THRESHOLD_VALUE = 230 # 二值化时的阈值PRETREATMENT_FILE = 'hq' # 横切时临时保存的文件夹W = 540 # 最小宽度H = 960 # 最小高度 class Pretreatment(object): __doc__ = "图片横向切割" def __init__(self, path, save_path, min_size=960): self.x = 0 self.y = 0 self.img_section = [] self.continuity_position = [] self.path = path self.save_path = save_path self.img_obj = None self.min_size = min_size self.mkdir(self.save_path) self.file_name = self.path.split('/')[-1] def get_continuity_position_new(self): img = cv2.imread(self.path) gray_image = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) ret, thresh1 = cv2.threshold(gray_image, THRESHOLD_VALUE, 255, cv2.THRESH_BINARY) width = img.shape[1] height = img.shape[0] self.x = width self.y = height for i in range(0, height): if thresh1[i].sum() != 255 * width: self.continuity_position.append(i) def filter_rule(self): if self.y < self.min_size: return True def mkdir(self, path): if not os.path.exists(path): os.makedirs(path) def get_section(self): # 获取区间 for k, g in groupby(enumerate(self.continuity_position), lambda x: x[1] - x[0]): l1 = [j for i, j in g] # 连续数字的列表 if len(l1) > 1: self.img_section.append([min(l1), max(l1)]) def split_img(self): print(self.img_section) for k, s in enumerate(self.img_section): if s: if not self.img_obj: self.img_obj = Image.open(self.path) if self.x < W: return if s[1] - s[0] < H: return cropped = self.img_obj.crop((0, s[0], self.x, s[1])) # (left, upper, right, lower) self.mkdir(os.path.join(self.save_path, PRETREATMENT_FILE)) cropped.save(os.path.join(self.save_path, PRETREATMENT_FILE, f"hq_{k}_{self.file_name}")) def remove_raw_data(self): os.remove(self.path) def main(self): # v2 try: self.get_continuity_position_new() self.filter_rule() self.get_section() self.split_img() except Exception as e: print(self.file_name) print(e) finally: if self.img_obj: self.img_obj.close() class Longitudinal(Pretreatment): def get_continuity_position_new(self): print(self.path) img = cv2.imread(self.path) gray_image = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) ret, thresh1 = cv2.threshold(gray_image, THRESHOLD_VALUE, 255, cv2.THRESH_BINARY) width = img.shape[1] height = img.shape[0] print(width, height) self.x = width self.y = height for i in range(0, width): if thresh1[:, i].sum() != 255 * height: self.continuity_position.append(i) def split_img(self): print(self.img_section) for k, s in enumerate(self.img_section): if s: if not self.img_obj: self.img_obj = Image.open(self.path) if self.y < H: return if s[1] - s[0] < W: return cropped = self.img_obj.crop((s[0], 0, s[1], self.y)) # (left, upper, right, lower) cropped.save(os.path.join(self.save_path, f"{k}_{self.file_name}")) def main(path, save_path): starttime = datetime.datetime.now() a = Pretreatment(path=path, save_path=save_path) a.main() for root, dirs, files in os.walk(os.path.join(save_path, PRETREATMENT_FILE)): for i in files: b = Longitudinal(path=os.path.join(save_path, PRETREATMENT_FILE, i), save_path=save_path) b.main() os.remove(os.path.join(save_path, PRETREATMENT_FILE, i)) endtime = datetime.datetime.now() print(f'耗时:{(endtime - starttime)}') if __name__ == '__main__': path = '你图片存放的路径' save_path = '要保存的路径' for _, _, files in os.walk(path): for i in files: main(path=os.path.join(path, i), save_path=save_path) os.rmdir(os.path.join(save_path, PRETREATMENT_FILE))
