File size: 1,746 Bytes
25ab7f8
 
 
654f874
25ab7f8
934a274
25ab7f8
 
 
 
 
 
 
934a274
 
 
 
25ab7f8
934a274
25ab7f8
 
 
2a0f77b
25ab7f8
 
 
1bcb537
 
25ab7f8
934a274
 
25ab7f8
 
 
 
934a274
 
2112a66
934a274
 
 
 
 
 
 
 
 
 
 
25ab7f8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
import cv2
from skimage import feature
import matplotlib.pyplot as plt
import numpy as np


class LBPImageEncoder:
    def __init__(self, numPoints, radius):
        self.numPoints = numPoints
        self.radius = radius

    def describe(self, image, eps=1e-7):
        lbp = feature.local_binary_pattern(image, self.numPoints, self.radius)
        hist, _ = np.histogram(lbp.ravel(), bins=np.arange(0, self.numPoints + 3), range=(0, self.numPoints + 2))
        hist = hist.astype("float")
        hist /= (hist.sum() + eps)
        return hist

    def face_detection(self, image):
        cascadePath = "haarcascade_frontalface_default.xml"
        detector = cv2.CascadeClassifier(cascadePath)

        gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
        rects = detector.detectMultiScale(gray, scaleFactor=1.05, minNeighbors=10, minSize=(30, 30),flags=cv2.CASCADE_SCALE_IMAGE)
        return rects

    def preprocess_img(self, image):
        img = np.array(image)
        rects = self.face_detection(img)
        feature_vector = np.zeros((self.numPoints + 2) * 3)

        for (x, y, w, h) in rects:
            face = img[y:y + h, x:x + w]
            face = cv2.cvtColor(face, cv2.COLOR_BGR2GRAY)

            lbp = self.describe(face)
            feature_vector[(self.numPoints + 2) * 0: (self.numPoints + 2) * 1] += lbp

            # Process green channel
            lbp = self.describe(face[:, :, 1])
            feature_vector[(self.numPoints + 2) * 1: (self.numPoints + 2) * 2] += lbp

            # Process blue channel
            lbp = self.describe(face[:, :, 2])
            feature_vector[(self.numPoints + 2) * 2: (self.numPoints + 2) * 3] += lbp

        feature_vector /= len(rects)

        return feature_vector