SIFT特征提取

opencv4.4版本由于sift专利原因已经无法使用，opencv4.2与python3.8不匹配，python版本需降到3.6，opencv版本需降到3.4.2.17

pip uninstall opencv-python
pip install opencv-python==3.4.2.17
pip install opencv-contrib-python==3.4.2.17
import pandas as pd
import pickle

提取图像中的特征点，并在原图像中显示出来

import cv2
img_path = r'../image/paojie.jpg'
img = cv2.imread(img_path)
# print(img.shape)
# img = cv2.resize(img,(136 * 3,512 * 3))
cv2.imshow("original", img)

gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)

# 使用SIFT
sift = cv2.xfeatures2d.SIFT_create()
keypoints, descriptor = sift.detectAndCompute(gray, None)

cv2.drawKeypoints(image=img,
                  outImage=img,
                  keypoints=keypoints,
                  flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS,
                  color=(51, 163, 236))
cv2.imshow("SIFT", img)

img = cv2.imread(img_path)
# img = cv2.resize(img,(136 * 3,76 * 3))

while True:
    if cv2.waitKey() & 0xff == ord('q'):
        break
cv2.destroyAllWindows()

提取图像中的SIFT特征点，并利用PCA(主成分分析)进行降维，并提取特征值。注：SIFT提取出来的特征是一个128维的矩阵，我在这里利用PCA主成分分析将矩阵降为100维。

import cv2
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler

img_path = r'../image/paojie.jpg'
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
# 使用SIFT
sift = cv2.xfeatures2d.SIFT_create()
keypoints, descriptor = sift.detectAndCompute(gray, None)
# print(descriptor.shape)
# m,n=descriptor.shape
# print(m,n)
descriptor = StandardScaler().fit_transform(descriptor)
pca = PCA(n_components=100)
pca.fit(descriptor)
print(pca.singular_values_)  # 查看特征值
print(pca.components_)  # 打印查看特征值对应的特征向量
# print(pca.components_.shape)

cv2.drawKeypoints(image=img,
                  outImage=img,
                  keypoints=keypoints,
                  flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS,
                  color=(255, 0, 255))
cv2.imshow("SIFT", img)

while True:
    if cv2.waitKey() & 0xff == ord('q'):
        break
cv2.destroyAllWindows()

flags不同，flags为绘制点的模式

import cv2

img = cv2.imread('../image/paojie.jpg', cv2.IMREAD_COLOR)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
cv2.imshow('origin', img);

# SIFT
detector = cv2.xfeatures2d.SIFT_create()
keypoints = detector.detect(gray, None)
img = cv2.drawKeypoints(image=gray,
                        keypoints=keypoints,
                        outImage=None,
                        color=(255, 0, 255))
# img = cv2.drawKeypoints(gray,keypoints,flags = cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
cv2.imshow('test', img)
while True:
    if cv2.waitKey() & 0xff == ord('q'):
        break
cv2.destroyAllWindows()

给定两张图片，计算其SIFT特征匹配结果

# coding: utf-8
from matplotlib import pyplot as plt
from imagedt.decorator import time_cost
import cv2

print('cv version: ', cv2.__version__)
def bgr_rgb(img):
    (r, g, b) = cv2.split(img)
    return cv2.merge([b, g, r])


def orb_detect(image_a, image_b):
    # feature match
    orb = cv2.ORB_create()

    kp1, des1 = orb.detectAndCompute(image_a, None)
    kp2, des2 = orb.detectAndCompute(image_b, None)

    # create BFMatcher object
    bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)

    # Match descriptors.
    matches = bf.match(des1, des2)

    # Sort them in the order of their distance.
    matches = sorted(matches, key=lambda x: x.distance)

    # Draw first 10 matches.
    img3 = cv2.drawMatches(image_a, kp1, image_b, kp2, matches[:100], None, flags=2)

    return bgr_rgb(img3)


@time_cost
def sift_detect(img1, img2, detector='surf'):
    if detector.startswith('si'):
        print("sift detector......")
        sift = cv2.xfeatures2d.SURF_create()
    else:
        print("surf detector......")
        sift = cv2.xfeatures2d.SURF_create()

    # find the keypoints and descriptors with SIFT
    kp1, des1 = sift.detectAndCompute(img1, None)
    kp2, des2 = sift.detectAndCompute(img2, None)

    # BFMatcher with default params
    bf = cv2.BFMatcher()
    matches = bf.knnMatch(des1, des2, k=2)

    # Apply ratio test
    good = [[m] for m, n in matches if m.distance < 0.5 * n.distance]

    # cv2.drawMatchesKnn expects list of lists as matches.
    img3 = cv2.drawMatchesKnn(img1, kp1, img2, kp2, good, None, flags=2)

    return bgr_rgb(img3)


if __name__ == "__main__":
    # load image
    image_a = cv2.imread('../image/dongman1.jpg')
    image_b = cv2.imread('../image/dongman2.jpg')

    # ORB
    # img = orb_detect(image_a, image_b)

    # SIFT or SURF
    img = sift_detect(image_a, image_b)

    plt.imshow(img)
    plt.show()

首先获取全部图片的特征数据

import cv2
import numpy as np
from os import walk
from os.path import join

def create_descriptors(folder):
    files = []
    for (dirpath, dirnames, filenames) in walk(folder):
        files.extend(filenames)
    for f in files:
        if '.jpg' in f:
            save_descriptor(folder, f, cv2.xfeatures2d.SIFT_create())

def save_descriptor(folder, image_path, feature_detector):
    # 判断图片是否为npy格式
    if image_path.endswith("npy"):
        return
    # 读取图片并检查特征
    img = cv2.imread(join(folder,image_path), 0)
    keypoints, descriptors = feature_detector.detectAndCompute(img, None)
    # 设置文件名并将特征数据保存到npy文件
    descriptor_file = image_path.replace("jpg", "npy")
    np.save(join(folder, descriptor_file), descriptors)

if __name__=='__main__':
    path = 'D://PycharmProjects//pythonProject//image'
    create_descriptors(path)

将图片的特征数据保存在npy文件。下一步是根据选择的图域这些特征数据文件进行匹配，从而找出最佳匹配的图片。

from os.path import join
from matplotlib import pyplot as plt
from os import walk
import numpy as np
import cv2

query = cv2.imread('D://PycharmProjects//pythonProject//image/1.jpg', 0)
folder = 'D:\\PycharmProjects\\pythonProject\\image'
descriptors = []
# 获取特征数据文件名
for (dirpath, dirnames, filenames) in walk(folder):
    for f in filenames:
        if f.endswith("npy"):
            descriptors.append(f)
    print(descriptors)

# 使用SIFT算法检查图像的关键点和描述符
sift = cv2.xfeatures2d.SIFT_create()
query_kp, query_ds = sift.detectAndCompute(query, None)

# 创建FLANN匹配器
index_params = dict(algorithm=0, trees=5)
search_params = dict(checks=50)
flann = cv2.FlannBasedMatcher(index_params, search_params)

potential_culprits = {}
for d in descriptors:
    # 将图像query与特征数据文件的数据进行匹配
    matches = flann.knnMatch(query_ds, np.load(join(folder, d)), k=2)
    # 清除错误匹配
    good = []
    for m, n in matches:
        if m.distance < 0.7 * n.distance:
            good.append(m)
    # 输出每张图片与目标图片的匹配数目
    print("img is %s ! matching rate is (%d)" % (d, len(good)))
    potential_culprits[d] = len(good)

max_matches = None
potential_suspect = None
for culprit, matches in potential_culprits.items():
    if max_matches == None or matches > max_matches:
        max_matches = matches
        potential_suspect = culprit

print("potential suspect is %s" % potential_suspect.replace("npy", "").upper())

高斯混合建模、背景差分提取前景目标，并显示轮廓标示

import numpy as np
import cv2
class demo2():
    def __init__(self, Videopath='../video/李永乐老师.mp4'):
        self.capture = cv2.VideoCapture(Videopath)

    def Gaussian(self, drawContours=False, drawRectangle=True):
        cap = self.capture
        # 创建形态学操作时需要使用的核
        kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
        # 创建混合高斯模型
        fgbg = cv2.createBackgroundSubtractorMOG2()
        # 将行人在视频中实时标记出
        while (True):
            ret, frame = cap.read()
            fgmask = fgbg.apply(frame)
            # 形态学开运算去噪点
            fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_OPEN, kernel)

            # 寻找视频中的轮廓
            im, contours, hierarchy = cv2.findContours(fgmask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

            if drawContours:  # 背景差分提取前景目标，将轮廓信息标示
                n = len(contours)
                for i in range(n):
                    temp = np.zeros(frame.shape, np.uint8)
                    temp = cv2.drawContours(temp, contours, i, (255, 255, 255), 2)
                    cv2.imshow('frame', frame)
                    cv2.imshow("contours", temp)
                cv2.waitKey()

            if drawRectangle:
                for c in contours:
                    # 计算各轮廓的周长
                    perimeter = cv2.arcLength(c, True)
                    if perimeter > 188:
                        # 找到一个直矩形（不会旋转）
                        x, y, w, h = cv2.boundingRect(c)
                        # 画出这个矩形
                        cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)

                cv2.imshow('frame', frame)
                cv2.imshow('fgmask', fgmask)
                k = cv2.waitKey(20)
                if k == 27:
                    break

        cap.release()
        cv2.destroyAllWindows()
demo2().Gaussian()

import cv2
def demo3():
    # 初始化视频捕获设备
    # gVideoDevice = cv2.VideoCapture("./video.avi")
    gVideoDevice = cv2.VideoCapture(0)
    if not gVideoDevice.isOpened():
        print('open video failed')
        return
    else:
        print('open video succeeded')

    # 选择 框选帧
    print("按 enter 选择当前帧，否则继续下一帧")
    while True:
        gCapStatus, gFrame = gVideoDevice.read()
        cv2.imshow("pick frame", gFrame)
        k = cv2.waitKey()
        if k == 13:
            break

    # 框选感兴趣区域
    cv2.destroyWindow("pick frame")
    gROI = cv2.selectROI("ROI frame", gFrame, False)
    if (not gROI):
        print("空框选，退出")
        quit()

    # 初始化追踪器
    gTracker = cv2.TrackerKCF_create()
    gTracker.init(gFrame, gROI)

    # 循环帧读取，开始跟踪
    while True:
        gCapStatus, gFrame = gVideoDevice.read()
        if (gCapStatus):
            # 展示跟踪图片
            status, coord = gTracker.update(gFrame)
            if status:
                message = {"coord": [((int(coord[0]), int(coord[1])),
                                      (int(coord[0] + coord[2]), int(coord[1] + coord[3])))]}
                p1 = (int(coord[0]), int(coord[1]))
                p2 = (int(coord[0] + coord[2]), int(coord[1] + coord[3]))
                cv2.rectangle(gFrame, p1, p2, (255, 0, 0), 2, 1)
                message['msg'] = "is tracking"
            else:
                message['msg'] = "KCF error,需要重新使用调用跟踪器"
            cv2.imshow('tracked image', gFrame)
            print(message)
            key = cv2.waitKey()
            if key == 27:
                break
        else:
            print("捕获帧失败")
            quit()

demo3()