FaceDect/SimpleClassifier.h at master · Lyusijie/FaceDect · GitHub

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
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
#ifndef SIM_H
#define SIM_H
#include "IntImage.h"
#include <fstream>
#include "MyStruct.h"
#include "ocv.h"

struct SimpleClassifier
{
	REAL thresh;
	REAL error;
	int parity;
	int type; // which type of feature?
	int x1,x2,x3,x4,y1,y2,y3,y4;

	inline const REAL GetOneFeature(const IntImage& im) const;
	inline const REAL GetOneFeatureTranslation(REAL** const data,const int y) const;
	inline const int Apply(const REAL value) const;
	inline const int Apply(const IntImage& im) const;
	void WriteToFile(std::ofstream& f) const;
	void ReadFromFile(std::ifstream& f);
};

const int SimpleClassifier::Apply(const IntImage& im) const
{
	if(parity == 1)
		return (GetOneFeature(im)<thresh)?1:0;
	else
		return (GetOneFeature(im)>=thresh)?1:0;
}

const int SimpleClassifier::Apply(const REAL value) const
{
	if(parity == 1)
		return (value<thresh)?1:0;
	else
		return (value>=thresh)?1:0;
}

const REAL SimpleClassifier::GetOneFeature(const IntImage& im) const
{
	REAL f1;
	REAL** data = im.data;

	switch(type)
	{
	case 0:
		f1 =   data[x1][y3] - data[x1][y1] + data[x3][y3] - data[x3][y1]
			 + 2*(data[x2][y1] - data[x2][y3]);
		break;
	case 1:
		f1 =   data[x3][y1] + data[x3][y3] - data[x1][y1] - data[x1][y3]
			 + 2*(data[x1][y2] - data[x3][y2]);
		break;
	case 2:
		f1 =   data[x1][y1] -data[x1][y3] + data[x4][y3] - data[x4][y1]
			 + 3*(data[x2][y3] - data[x2][y1] + data[x3][y1] - data[x3][y3]);
		break;
	case 3:
		f1 =   data[x1][y1] - data[x1][y4] + data[x3][y4] - data[x3][y1]
			 + 3*(data[x3][y2] - data[x3][y3] + data[x1][y3] - data[x1][y2] );
		break;
	case 4:
		f1 =   data[x1][y1] + data[x1][y3] + data[x3][y1] + data[x3][y3]
			 - 2*(data[x2][y1] + data[x2][y3] + data[x1][y2] + data[x3][y2])
			 + 4*data[x2][y2];
		break;
	}
	return f1/im.variance;
}

const REAL SimpleClassifier::GetOneFeatureTranslation(REAL** const data,const int y) const
{
	REAL f1=0;

	switch(type)
	{
	case 0:
		f1 =   data[x1][y+y3] - data[x1][y+y1] + data[x3][y+y3] - data[x3][y+y1]
			 + 2*(data[x2][y+y1] - data[x2][y+y3]);
		break;
	case 1:
		f1 =   data[x3][y+y1] + data[x3][y+y3] - data[x1][y+y1] - data[x1][y+y3]
			 + 2*(data[x1][y+y2] - data[x3][y+y2]);
		break;
	case 2:
		f1 =   data[x1][y+y1] - data[x1][y+y3] + data[x4][y+y3] - data[x4][y+y1]
			 + 3*(data[x2][y+y3] - data[x2][y+y1] + data[x3][y+y1]  - data[x3][y+y3]);
		break;
	case 3:
		f1 =   data[x1][y+y1] - data[x1][y+y4] + data[x3][y+y4] - data[x3][y+y1]
			 + 3*(data[x3][y+y2] - data[x3][y+y3] + data[x1][y+y3] - data[x1][y+y2]);
		break;
	case 4:
		f1 =   data[x1][y+y1] + data[x1][y+y3] + data[x3][y+y1] + data[x3][y+y3]
			 - 2*(data[x2][y+y1] + data[x2][y+y3] + data[x1][y+y2] + data[x3][y+y2])
			 + 4*data[x2][y+y2];
		break;
	default:
		break;
	}
	return f1;
}
#endif