myOCR/image_tools.py

# -*- coding:utf-8 -*-
import math
import numpy as np
# from neighbor import neighbor_change
from Near import neighbor_change
import numba as nb

def get_range(bboxs):
    # 坐标转化
    ranges = []
    for i in bboxs:
        xs = i[::2]
        ys = i[1::2]
        box = [min(xs), min(ys), max(xs), max(ys)]
        ranges.append(box)
    return ranges

# @nb.jit
def check_range(pixel_points, ranges):
    """

    :param pixel_points: 一团联通坐标
    :param ranges:  所有textbox坐标
    :return:
    """
    pixel_point = pixel_points[0]
    for i in ranges:
        if i[2] >= pixel_point[0] >= i[0] and i[3] >= pixel_point[1] >= i[1]:
            return 1
    return 0


def is_text(bboxs, contours):
    '''
    doc
    判断一个实物点在不在textbox中
    :param bboxs: text的区域
    :param contours: 所有区域,[array[一个x坐标，一个y坐标]]
    :return: 插图区域
    '''
    ranges = bboxs
    illustration_box = []
    for region in contours:
        contain = [check_range(pixel_points, ranges) for pixel_points in region]
        contain = sum(contain) / len(contain)
        if contain < 0.2:  # 交并比 iou

            #  坐标结构化
            pixel = [math.inf, math.inf, -1, -1]
            for i in region:
                if i[0][0] < pixel[0]:
                    pixel[0] = i[0][0]
                if i[0][0] > pixel[2]:
                    pixel[2] = i[0][0]
                if i[0][1] < pixel[1]:
                    pixel[1] = i[0][1]
                if i[0][1] > pixel[3]:
                    pixel[3] = i[0][1]

            x_min, y_min, x_max, y_max = pixel
            if (x_max - x_min) * (y_max - y_min) > 30:  # 30是一个参数  像素点超过30
                illustration_box.append(pixel)
    return illustration_box


def processed(text_boxes, i_box):
    # 去除大图片中的小图片
    x_min, y_min, x_max, y_max = 0, 1, 2, 3  # index
    if len(i_box):
        W = i_box[-1][2] - i_box[-1][0] * 0.9
        i_box = [i for i in i_box if i[2] - i[0] < W]

    tmp_bbx = []
    for i in range(len(i_box)):
        for j in range(len(i_box)):
            if i_box[i][x_min] > i_box[j][x_min] and i_box[i][y_min] > i_box[j][y_min] \
                    and i_box[i][x_max] < i_box[j][x_max] and i_box[i][y_max] < i_box[j][y_max]:
                tmp_bbx.append(i)

    inx = [i for i in range(len(i_box)) if i not in tmp_bbx]

    i_box_c = [i_box[i] for i in inx]
    text_boxes, i_box_c = neighbor_change(text_boxes, i_box_c)
    return i_box_c, text_boxes