目录
1.学习目标
2.图像特征理解
2.1图像特征
2.2颜色特征
2.3纹理特征
2.4形状特征
2.5空间关系特征
3.形状特征描述
3.1HOG特征提取
3.2HOG实现过程
3.3角点概念
3.4Harris角点检测
3.5Harris实现过程
3.6SIFT算法
3.7SIFT实现过程
4.LBP纹理特征
4.1LBP介绍
4.2LBP原理
5.模板匹配算法
6.人脸检测算法
6.1人脸检测
6.2人脸对齐
6.3人脸特征提取
6.4人脸识别
第五点补充说明:将0-360度分布的梯度方向映射到bin个小的区间内。
代码:
import cv2
import numpy as np
# 判断矩形i是否完全包含在矩形o中
def is_inside(o, i):
ox, oy, ow, oh = o
ix, iy, iw, ih = i
return ox > ix and oy > iy and ox + ow < ix + iw and oy + oh < iy + ih
# 对人体绘制颜色框
def draw_person(image, person):
x, y, w, h = person
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 255), 2)
img = cv2.imread("people.jpg")
hog = cv2.HOGDescriptor() # 启动检测器对象
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector()) # 指定检测器类型为人体
found, w = hog.detectMultiScale(img,0.1,(1,1)) # 加载并检测图像
print(found)
print(w)
# 丢弃某些完全被其它矩形包含在内的矩形
found_filtered = []
for ri, r in enumerate(found):
for qi, q in enumerate(found):
if ri != qi and is_inside(r, q):
break
else:
found_filtered.append(r)
print(found_filtered)
# 对不包含在内的有效矩形进行颜色框定
for person in found_filtered:
draw_person(img, person)
cv2.imshow("people detection", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出:
[[322 122 92 184]
[228 107 94 188]
[ 97 109 92 185]
[394 132 87 173]
[157 143 89 178]]
[[1.69752147]
[1.50364332]
[2.26940114]
[0.15458478]
[1.37897637]]
[array([322, 122, 92, 184], dtype=int32)]
[array([322, 122, 92, 184], dtype=int32), array([228, 107, 94, 188], dtype=int32)]
[array([322, 122, 92, 184], dtype=int32), array([228, 107, 94, 188], dtype=int32), array([ 97, 109, 92, 185], dtype=int32)]
[array([322, 122, 92, 184], dtype=int32), array([228, 107, 94, 188], dtype=int32), array([ 97, 109, 92, 185], dtype=int32), array([394, 132, 87, 173], dtype=int32)]
[array([322, 122, 92, 184], dtype=int32), array([228, 107, 94, 188], dtype=int32), array([ 97, 109, 92, 185], dtype=int32), array([394, 132, 87, 173], dtype=int32), array([157, 143, 89, 178], dtype=int32)]
import cv2
import numpy as np
filename = 'harris2.png'
img = cv2.imread(filename)
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
gray = np.float32(gray)
# 输入图像必须是 float32 ,最后一个参数在 0.04 到 0.06 之间
dst = cv2.cornerHarris(gray,2,3,0.06)
#结果进行膨胀,可有可无
dst = cv2.dilate(dst,None)
print(dst)
# 设定阈值,不同图像阈值不同
img[dst>0.01*dst.max()]=[0,0,255]
print(dst.max())
cv2.imshow('dst_img',img)
#cv2.imshow('dst',dst)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出:
[[0. 0. 0. ... 0. 0. 0.]
[0. 0. 0. ... 0. 0. 0.]
[0. 0. 0. ... 0. 0. 0.]
...
[0. 0. 0. ... 0. 0. 0.]
[0. 0. 0. ... 0. 0. 0.]
[0. 0. 0. ... 0. 0. 0.]]
533091900.0
import cv2
import numpy as np
img = cv2.imread('harris2.png')
gray= cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
sift = cv2.xfeatures2d.SIFT_create()
kp = sift.detect(gray,None)#找到关键点
img=cv2.drawKeypoints(gray,kp,img)#绘制关键点
cv2.imshow('sp',img)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出:
def LBP(src):
'''
:param src:灰度图像
:return:
'''
height = src.shape[0]
width = src.shape[1]
dst = src.copy()
lbp_value = np.zeros((1,8), dtype=np.uint8)
#print(lbp_value)
neighbours = np.zeros((1,8), dtype=np.uint8)
#print(neighbours)
for x in range(1, width-1):
for y in range(1, height-1):
neighbours[0, 0] = src[y - 1, x - 1]
neighbours[0, 1] = src[y - 1, x]
neighbours[0, 2] = src[y - 1, x + 1]
neighbours[0, 3] = src[y, x - 1]
neighbours[0, 4] = src[y, x + 1]
neighbours[0, 5] = src[y + 1, x - 1]
neighbours[0, 6] = src[y + 1, x]
neighbours[0, 7] = src[y + 1, x + 1]
center = src[y, x]
for i in range(8):
if neighbours[0, i] > center:
lbp_value[0, i] = 1
else:
lbp_value[0, i] = 0
lbp = lbp_value[0, 0] * 1 + lbp_value[0, 1] * 2 + lbp_value[0, 2] * 4 + lbp_value[0, 3] * 8 \
+ lbp_value[0, 4] * 16 + lbp_value[0, 5] * 32 + lbp_value[0, 6] * 64 + lbp_value[0, 7] * 128
#print(lbp)
dst[y, x] = lbp
return dst
import cv2
import numpy as np
img = cv2.imread('people.jpg',0)
print(img.shape)
cv2.imshow('src',img)
cv2.waitKey(0)
cv2.destroyAllWindows()
new_img = LBP(img)
cv2.imshow('dst',new_img)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出:
(334, 500)
平方差匹配的方法速度最快,效果无法保证;相关性匹配的方法速度比平方差匹配慢,但效果要好一些;相关性系数匹配速度最慢,效果最好。
代码:
#模板匹配
import cv2
import numpy as np
def template_demo(tpl,target):
methods = [cv2.TM_SQDIFF_NORMED, cv2.TM_CCORR_NORMED, cv2.TM_CCOEFF_NORMED] #3种模板匹配方法
th, tw = tpl.shape[:2]
for md in methods:
#print(md)
result = cv2.matchTemplate(target, tpl, md)
#print(result.shape)
min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(result)
print(min_val, max_val, min_loc, max_loc)
if md == cv2.TM_SQDIFF_NORMED:
tl = min_loc
else:
tl = max_loc
br = (tl[0]+tw, tl[1]+th) #br是矩形右下角的点的坐标
cv2.rectangle(target, tl, br, (0, 0, 255), 2)
cv2.namedWindow("match-" + np.str(md), cv2.WINDOW_NORMAL)
cv2.imshow("match-" + np.str(md), target)
tpl =cv2.imread("sample2.jpg")
print(tpl.shape)
target = cv2.imread("target1.jpg")
print(target.shape)
cv2.waitKey(0)
cv2.destroyAllWindows()
cv2.namedWindow('template image', cv2.WINDOW_NORMAL)
cv2.imshow("template image", tpl)
cv2.namedWindow('target image', cv2.WINDOW_NORMAL)
cv2.imshow("target image", target)
template_demo(tpl,target)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出:
import cv2
# 读入图像
img = cv2.imread("12.jpg")
img = cv2.resize(img,None,fx=0.4,fy=0.4,interpolation=cv2.INTER_LINEAR)
# 加载人脸特征,该文件在 python安装目录\Lib\site-packages\cv2\data 下
face_cascade = cv2.CascadeClassifier(r'haarcascade_frontalface_default.xml')
# 将读取的图像转为COLOR_BGR2GRAY,减少计算强度
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# 检测出的人脸个数
faces = face_cascade.detectMultiScale(gray, scaleFactor = 1.15, minNeighbors = 4, minSize = (5, 5))
print("Face : {0}".format(len(faces)))
print(faces)
# 用矩形圈出人脸的位置
for(x, y, w, h) in faces:
cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.namedWindow("Faces")
cv2.imshow("Faces", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出:
Face : 5
[[128 110 48 48]
[188 107 50 50]
[338 81 54 54]
[260 93 53 53]
[ 55 106 44 44]]
代码2:
# -*- coding:utf-8 -*-
import cv2
import dlib
import numpy as np
predictor_model = 'shape_predictor_68_face_landmarks/shape_predictor_68_face_landmarks.dat'
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(predictor_model)
# cv2读取图像
test_film_path = "3.png"
img = cv2.imread(test_film_path)
# 取灰度
img_gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
# 人脸数rects
rects = detector(img_gray, 0)
print(rects[0])
for i in range(len(rects)):
landmarks = np.matrix([[p.x, p.y] for p in predictor(img,rects[i]).parts()])
print(landmarks, type(landmarks))
for idx, point in enumerate(landmarks):
# 68点的坐标
pos = (point[0, 0], point[0, 1])
#print(idx+1, pos)
# 利用cv2.circle给每个特征点画一个圈,共68个
cv2.circle(img, pos, 3, color=(0, 255, 0))
# 利用cv2.putText输出1-68
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(img, str(idx+1), pos, font, 0.5, (0, 0, 255), 1, cv2.LINE_AA)
#cv2.imwrite("result.png", img)
cv2.imshow("img", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
输出: