111 lines
3.5 KiB
Python
111 lines
3.5 KiB
Python
#!/usr/bin/env python
|
||
# -*- encoding: utf-8 -*-
|
||
'''
|
||
@File : qr_orb.py
|
||
@Contact : zpyovo@hotmail.com
|
||
@License : (C)Copyright 2018-2019, Lab501-TransferLearning-SCUT
|
||
@Description :
|
||
|
||
@Modify Time @Author @Version @Desciption
|
||
------------ ------- -------- -----------
|
||
2023/8/26 18:58 Pengyu Zhang 1.0 None
|
||
'''
|
||
|
||
from __future__ import absolute_import
|
||
from __future__ import division
|
||
from __future__ import print_function
|
||
|
||
import cv2
|
||
import numpy as np
|
||
|
||
def preprocess_image(image):
|
||
# 将图像调整为固定大小
|
||
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
||
resized_image = cv2.resize(image, (275, 275), interpolation=cv2.INTER_NEAREST)
|
||
return resized_image
|
||
|
||
def FAST_corner_detection(image, threshold):
|
||
keypoints = []
|
||
for i in range(3, image.shape[0] - 3):
|
||
for j in range(3, image.shape[1] - 3):
|
||
center_pixel = image[i, j]
|
||
pixel_difference = [abs(image[i + dx, j + dy] - center_pixel) for dx, dy in
|
||
[(0, -3), (0, 3), (-3, 0), (3, 0)]]
|
||
if all(diff > threshold for diff in pixel_difference):
|
||
keypoints.append(cv2.KeyPoint(j, i, 7))
|
||
return keypoints
|
||
|
||
|
||
def BRIEF_descriptor(keypoints, image):
|
||
patch_size = 31
|
||
descriptors = []
|
||
for kp in keypoints:
|
||
x, y = int(kp.pt[0]), int(kp.pt[1])
|
||
patch = image[y - patch_size // 2:y + patch_size // 2 + 1, x - patch_size // 2:x + patch_size // 2 + 1]
|
||
descriptor = ""
|
||
for i in range(patch_size * patch_size):
|
||
x1, y1 = np.random.randint(0, patch_size, size=2)
|
||
x2, y2 = np.random.randint(0, patch_size, size=2)
|
||
if patch[y1, x1] < patch[y2, x2]:
|
||
descriptor += "0"
|
||
else:
|
||
descriptor += "1"
|
||
descriptors.append(descriptor)
|
||
return descriptors
|
||
|
||
|
||
def ORB(image, nfeatures=500, fastThreshold=20):
|
||
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
||
|
||
keypoints = FAST_corner_detection(gray, fastThreshold)
|
||
keypoints = sorted(keypoints, key=lambda x: -x.response)[:nfeatures]
|
||
|
||
descriptors = BRIEF_descriptor(keypoints, gray)
|
||
|
||
return keypoints, descriptors
|
||
|
||
def ORB_detect(img,nfeatures=800,fastThreshold=20):
|
||
img = preprocess_image(img)
|
||
# 初始化opencv原生ORB检测器
|
||
orb = cv2.ORB_create(nfeatures=nfeatures, scaleFactor=1.2, edgeThreshold=31, patchSize=31, fastThreshold=fastThreshold)
|
||
|
||
kp, des = orb.detectAndCompute(img, None)
|
||
return des
|
||
|
||
|
||
'''
|
||
必要传参：
|
||
std_roi_feature: 服务器备案特征数据
|
||
ter_roi_feature: 手机终端特征数据
|
||
threshold: 相似度对比阈值
|
||
预留传参：
|
||
distance: 距离筛选度量，默认值100
|
||
'''
|
||
|
||
def roi_siml(std_roi_feature,ter_roi_feature,distance=100, threshold = None):
|
||
|
||
std_roi_feature = std_roi_feature.astype('uint8')
|
||
ter_roi_feature = ter_roi_feature.astype('uint8')
|
||
threshold = float(threshold)
|
||
|
||
# 使用汉明距离对特侦点距离进行计算
|
||
bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=False)
|
||
# 使用knn算法进行匹配
|
||
matches = bf.knnMatch(std_roi_feature, trainDescriptors=ter_roi_feature, k=2)
|
||
|
||
# 去除无效和模糊的匹配
|
||
good = []
|
||
for match in matches:
|
||
if len(match) >= 2:
|
||
good = [(m, n) for (m, n) in matches if m.distance < 0.95 * n.distance and m.distance < distance]
|
||
|
||
similarity = len(good) / len(matches) *100
|
||
if similarity >=threshold:
|
||
return 'passed',similarity
|
||
else:
|
||
return 'failed ',similarity
|
||
|
||
|
||
|
||
|