-
Notifications
You must be signed in to change notification settings - Fork 2
/
vibe.m
215 lines (210 loc) · 8.26 KB
/
vibe.m
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
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
clear all;
%算法参数初始化，N表示2D欧式坐标图像中，每个像素点建立的背景采样个数；R表示欧式距离阀值；每个像素
%
N=20;
R=25;
R1=25;
MIN=3;
Q=5;
lambda=0.25;
xyloObj = VideoReader('office.avi');
nFrames = xyloObj.NumberOfFrames;
vidHeight = xyloObj.Height;
vidWidth = xyloObj.Width;
mov(1:nFrames) = ...
struct('cdata', zeros(vidHeight, vidWidth, 3, 'uint8'),...
'colormap', []);
% Preallocate movie structure.用于存放灰度视频帧
image(1:nFrames) = ...
struct('cdata', zeros(vidHeight, vidWidth, 1, 'uint8'),...
'colormap', []);
for k = 1 : nFrames
mov(k).cdata = read(xyloObj, k);
image(k).cdata = rgb2gray(mov(k).cdata );
end
samples=zeros(vidHeight, vidWidth, N, 'uint8');
segMap = zeros(vidHeight, vidWidth, 1, 'uint8');
image0 = zeros(vidHeight, vidWidth, 'uint8');
blink = zeros(vidHeight, vidWidth, 'uint8');
image_blink = zeros(vidHeight, vidWidth, nFrames ,'uint8');
foregroundMatchCount = zeros(vidHeight, vidWidth, 1, 'uint8');
foregroundGoals=zeros(vidHeight, vidWidth, 3);%记录前景目标的RGB值
background=0;
foreground=255;
%提取背景
for x = 1 : vidHeight
for y = 1 : vidWidth
for k = 1 : nFrames
n(k)=image(k).cdata(x,y);
end
image0(x,y) = median (double(n));
end
end
%初始化样本集
for k = 1 : N
for x=1:vidHeight
for y=1:vidWidth
xNG=getRandomNeighbrXCoordinate( x,vidHeight);
yNG=getRandomNeighbrYCoordinate( y,vidWidth);
samples(x,y,k)=image0(x,y);
% samples(x,y,k)=image(1).cdata(xNG,yNG);
end
end
end
for k = 1: nFrames
for x=1:vidHeight
for y=1:vidWidth
count=0;
index=1;
dist=0;
while((count<MIN)&&(index<=N))
dist=abs(double(mov(k).cdata(x,y))-double(samples(x,y,index)));
%检测闪烁点
if k>2
if(image_blink(x,y,k)==image_blink(x,y,k-1))
blink(x,y)=blink(x,y)-1;
else
blink(x,y)=blink(x,y)+15;
end
end
if(blink(x,y)>30)
if(R<40)
R=R1*(1+lambda);
end
else
if(R>20)
R=R1*(1-lambda);
end
end
if(dist<R)
count=count+1;
end
index=index+1;
end
if(count>=MIN)
foregroundMatchCount(x,y)=0;
segMap(x,y)=background;
foregroundGoals(x,y,:)=0;
rand=randi([1 Q],1,1);
if(rand==1)
rand=randi([1 N],1,1);
samples(x,y,rand)=image(k).cdata(x,y);
end
rand=randi([1 Q],1,1);
if(rand==1)
xNG=getRandomNeighbrXCoordinate( x,vidHeight);
yNG=getRandomNeighbrYCoordinate( y,vidWidth);
rand=randi([1 N],1,1);
samples(xNG,yNG,rand)=image(k).cdata(x,y);
end
else
%鬼影检测
Hnum=10;Lnum=5;Tb1=3;Tb2=1;Tb3=0.5;Bth=4;Tblve=0;
k_neighborhood=[struct('x',-1,'y',-1) struct('x',-1,'y',0) struct('x',-1,'y',1) struct('x',0,'y',-1)];
for i=1:4
if(x+k_neighborhood(i).x<=0 || y+k_neighborhood(i).y<=0 || y+k_neighborhood(i).y>vidWidth)
continue;
end
count=0;
index=1;
dist=0;
while((count<MIN)&&(index<=N))
dist=abs(double(mov(k).cdata(x,y))-double(samples(x+k_neighborhood(i).x,y+k_neighborhood(i).y,index)));
if(dist<R)
count=count+1;
end
index=index+1;
end
if(count>=0 && count<=Lnum)
Tblve=Tblve+Tb3;
end
if(count>=Lnum && count<Hnum)
Tblve=Tblve+Tb2;
end
if(count>=Hnum)
Tblve=Tblve+Tb1;
end
end
if(Tblve>=Bth)%当前点为鬼影点
foregroundMatchCount(x,y)=0;
segMap(x,y)=background;
foregroundGoals(x,y,:)=0;
for i=1:N
rand=randi([1 4],1,1);
while(x+k_neighborhood(rand).x<=0 || y+k_neighborhood(rand).y<=0 || y+k_neighborhood(rand).y>vidWidth)
rand=randi([1 4],1,1);
end
samples(x,y,i)=samples(x+k_neighborhood(rand).x,y+k_neighborhood(rand).y,randi([1 N],1,1));
end
else
%消除摄像机抖动导致的噪点
area_24=[struct('x',-2,'y',-2) struct('x',-2,'y',-1) struct('x',-2,'y',0) struct('x',-2,'y',1) struct('x',-2,'y',2)...
struct('x',-1,'y',-2) struct('x',-1,'y',-1) struct('x',-1,'y',0) struct('x',-1,'y',1) struct('x',-1,'y',2)...
struct('x',0,'y',-2) struct('x',0,'y',-1) struct('x',0,'y',1) struct('x',0,'y',2)...
struct('x',1,'y',-2) struct('x',1,'y',-1) struct('x',1,'y',0) struct('x',1,'y',1) struct('x',1,'y',2)...
struct('x',2,'y',-2) struct('x',2,'y',-1) struct('x',2,'y',0) struct('x',2,'y',1) struct('x',2,'y',2)];
temp_N_th=0;N_th=1;
for i=1:24
if(x+area_24(i).x>=1 && x+area_24(i).x<=vidHeight && y+area_24(i).y>=1 && y+area_24(i).y<=vidWidth && segMap(x+area_24(i).x,y+area_24(i).y)==foreground)
temp_N_th=temp_N_th+1;
end
end
if(temp_N_th>N_th || k==1)%真实前景像素点
foregroundMatchCount(x,y)= foregroundMatchCount(x,y)+1;
segMap(x,y)=foreground;
foregroundGoals(x,y,:)=mov(k).cdata(x,y,:);
if(foregroundMatchCount(x,y)>nFrames*0.3)
foregroundMatchCount(x,y)=0;
segMap(x,y)=background;
foregroundGoals(x,y,:)=0;
%立即更新背景模型当中两个样本点，防止鬼影点再次出现
for i=1:2
rand=randi([1 N],1,1);
samples(x,y,rand)=image(k).cdata(x,y);
end
end
else
foregroundMatchCount(x,y)=0;
segMap(x,y)=background;
foregroundGoals(x,y,:)=0;
for i=1:2
rand=randi([1 N],1,1);
samples(x,y,rand)=image(k).cdata(x,y);
end
end
end
end
end
end
%丢弃小空洞
L = bwlabel(segMap,8);
max_L=max(max(L));
for s = 1:max_L
[m,n]=find(L==s);
r=length(m);
if r<20
for i=1:r
L(m(i),n(i))=0;
foregroundGoals(m(i),n(i),:)=0;
end
else
for j=1:r
L(m(j),n(j))=255;
end
end
end
%形态学填补空洞
se = strel('disk',1);
L_new = imdilate (L,se);
figure(2);
clf;
subplot(1,3,1),imshow(mov(k).cdata);
subplot(1,3,2),imshow(uint8(L_new));
% subplot(1,3,2),imshow(uint8(segMap));
subplot(1,3,3),imshow(uint8(foregroundGoals));
%视频导出图像
imwrite(L_new,['.\figure\前景',num2str(k),'(灰度).jpg']);
imwrite(mov(k).cdata,['.\figure\原始帧',num2str(k),'.jpg']);
% imwrite(segMap,['.\figure\前景',num2str(k),'(灰度).jpg']);
% imwrite(uint8(foregroundGoals),['.\figure\前景',num2str(k),'(RGB).jpg']);
end