-
Notifications
You must be signed in to change notification settings - Fork 4
/
depth_kalman_kinectv2.m
336 lines (286 loc) · 15 KB
/
depth_kalman_kinectv2.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
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
%% Kalman KinectV2 track of righthand
%% Reset previous
imaqreset;
clear;
clc;
close all;
%% create color and depth kinect video input objects
depthVid = videoinput('kinect', 2, 'Depth_512x424'); %start depth video
colorVid = videoinput('kinect', 1, 'BGR_1920x1080'); %start color video
triggerconfig(depthVid,'manual'); %manual triger video
triggerconfig(colorVid,'manual'); %manual trigger video
counter = 1; % Counter to store data
depth_focal_length_x = 388.198; %https://www.researchgate.net/figure/Parameters-from-the-calibration-of-the-Kinect-v2-by-rounding_tbl4_321048476
depth_focal_length_y = 389.033; %https://www.researchgate.net/figure/Parameters-from-the-calibration-of-the-Kinect-v2-by-rounding_tbl4_321048476
color_focal_length_x = 1144.361; %https://www.researchgate.net/figure/Parameters-from-the-calibration-of-the-Kinect-v2-by-rounding_tbl4_321048476
color_focal_length_y = 1147.337; %https://www.researchgate.net/figure/Parameters-from-the-calibration-of-the-Kinect-v2-by-rounding_tbl4_321048476
temp_time = 0; % store time temporary for velocity and acceleratio calc
framesPerTrig = 1; % 1 frame per second
depthVid.FramesPerTrigger=framesPerTrig; % set the frame trig frequency
depthVid.TriggerRepeat=inf; %loop forever
colorVid.FramesPerTrigger=framesPerTrig; %set frame trig frequency
colorVid.TriggerRepeat=inf; %repeat trig for infinite time
colorVid.Timeout = 100; %wait sec to obtain frame
src = getselectedsource(depthVid); %get source of depth video
src.EnableBodyTracking = 'on'; %enable skeleton tracking
start(depthVid); %start depth Video
start(colorVid); %start color Video
himg = figure; %first frame initialize
%% Fix Skeleton Mapping
SkeletonConnectionMap = [ [4 3]; %Head - Neck
[3 21]; %Neck - SpineShoulder
[21 2]; %SpineShoulder - SpineMid
[2 1]; %SpineMid - SpineBase
[21 9]; %SpineShoulder - ShoulderRight
[9 10]; %ShoulderRight - ElbowRight
[10 11]; %ElbowRight - WristRight
[11 12]; %WristRight - HandWright
[12 24]; %HandRight - HandTipRight
[12 25]; %HandRight - ThumbRight
[21 5]; %SpineShoulder - ShoulderLeft
[5 6]; %ShoulderLeft - ElbowLeft
[6 7]; %ElbowLeft - WristLeft
[7 8]; %WristLeft - HandLeft
[8 22]; %HandLeft - HandTipLeft
[8 23]; %HandLeft - ThumbLeft
[1 17]; %SpineBase - HipRight
[17 18]; %HipRight - KneeRight
[18 19]; %KneeRight - AnkleRight
[19 20]; %AnkleRight - FootRight
[1 13]; %SpineBase - HipLeft
[13 14]; %Hip Left - KneeLeft
[14 15]; %KneeLeft - AnkleLeft
[15 16]; ]; %AnkleLeft - FootLeft
%% Initialize parameters for kalmam filter
flagStart = 0;
time(counter) = 0;
dt = 30; %Initialize refresh rate
frame_dt = 1/30;
Q_loc = []; %real path
vel = []; %real velocity
Q_loc_meas = []; %real path
u = .005;
noise_mag = 4; %noise for velocity how fast is speeding up
noise_x = 10; noise_y = 10; noise_z = 10; % rule of thumb ---> more noise is better
Ez = [noise_x 0 0; ...
0 noise_y 0; ...
0 0 noise_z]; % noise matrix for x y z position measurements
Ex = [dt^4/4 0 0 dt^3/2 0 0; ...
0 dt^4/4 0 0 dt^3/2 0; ...
0 0 dt^4/4 0 0 dt^3/2; ...
dt^3/2 0 0 dt^2 0 0; ...
0 dt^3/2 0 0 dt^2 0; ...
0 0 dt^3/2 0 0 dt^2].*noise_mag^2; %Convert process noise (stdv) into covariance matrix
P = Ex;% estimate of initial position variance (covariance matrix)
% xt+1 = xt + Vxt + 0.5*ax*t^2 for one dimension
% Vxt+1 = Vxt + axt
% [x] [x]
% [y] [y]
% [z] = A * [z] + B * u
% [Vx] [Vx]
% [Vy] [Vy]
% [Vz] [Vz]
% zt = C * [x;y;z;Vx;Vy;Vz]
A = [1 0 0 dt 0 0; ...
0 1 0 0 dt 0; ...
0 0 1 0 0 dt; ...
0 0 0 1 0 0; ...
0 0 0 0 1 0; ...
0 0 0 0 0 1]; %State matrix for position in 3D space
B = [dt^2/2; ...
dt^2/2; ...
dt^2/2; ...
dt; ...
dt; ...
dt]; %State matrix
C = [1 0 0 0 0 0; ...
0 1 0 0 0 0; ...
0 0 1 0 0 0]; %We multiply this to get the next predicted state only for position
Q_loc_estimate = [];
vel_estimate = []; %velocity estimation
P_estimate = P;
%% Start mapping skeleton
while ishandle(himg) %loop for every frame
trigger(depthVid); %start triggering depth stream 1 frame per second
trigger(colorVid); %start triggering color stream 1 frame per second
[depthMap, depthTime, depthMetaData] = getdata(depthVid); %get data from video
[colorMap] = getdata(colorVid); % get colordata
bodies = depthMetaData.IsBodyTracked; % check if body is tracked
trackedBodies = find(bodies); % find how many bodies are tracked
nBodies = length(trackedBodies); %find how many bodies are tracked
depthHeight = size(depthMap, 1); % find depth Height
depthWidth = size(depthMap ,2); % find depth Width
colors = ['r';'g';'b';'c';'y';'m']; % colors for different bodies
imshow(depthMap, [0 4096]); % display depth frame
if (sum(bodies)) > 0 % if body is tracked
imshow(depthMap, [0 4096]); % display depth frame
skeletonJoints = depthMetaData.DepthJointIndices(:,:,depthMetaData.IsBodyTracked); % get the skeleton joints
posJoints = depthMetaData.JointPositions(:,:,depthMetaData.IsBodyTracked); % get the real positions of tracked joints in meters
hand_state = depthMetaData.HandRightState(trackedBodies); % 0 unknown | 1 not tracked | 2 open | 3 closed | 4 lasso
hand_confidence = depthMetaData.HandRightConfidence(trackedBodies); % 0 Low Confidence | 1 High Confidence
if hand_confidence == 1
flagStart = 1;
end
% Plot hand states
if hand_state == 0
text('units','pixels','position',[100 100],'color','red','fontsize',20,'string','Unknown');
elseif hand_state == 1
text('units','pixels','position',[100 100],'color','red','fontsize',20,'string','Not Tracked');
elseif hand_state == 2
text('units','pixels','position',[100 100],'color','red','fontsize',20,'string','Open');
elseif hand_state == 3
text('units','pixels','position',[100 100],'color','red','fontsize',20,'string','Closed');
else
text('units','pixels','position',[100 100],'color','red','fontsize',20,'string','Lasso')
end
if hand_confidence == 0
text('units','pixels','position',[100 130],'color','red','fontsize',20,'string','Low Confidence');
else
text('units','pixels','position',[100 130],'color','red','fontsize',20,'string','High Confidence');
end
hold on; %hold figure
for i = 8:9 % only for joints at the 9 and 10 line of skeleton connection map
for body = 1:nBodies % for every body tracked
% Coordinates of skeleton joints in depth space
X1 = [skeletonJoints(SkeletonConnectionMap(i,1),1,body); skeletonJoints(SkeletonConnectionMap(i,2),1,body)];
Y1 = [skeletonJoints(SkeletonConnectionMap(i,1),2,body); skeletonJoints(SkeletonConnectionMap(i,2),2,body)];
% Real world coordinates of joints
hand_pos_x = posJoints(12,1,body); %meters
hand_pos_y = posJoints(12,2,body); %meters
hand_pos_z = posJoints(12,3,body); %meters
thumb_x = posJoints(25,1,body); %meters
thumb_y = posJoints(25,2,body); %meters
thumb_z = posJoints(25,3,body); %meters
% Calculate orientation based on derivative of line
% between arm and hand
line_v = [1e3*(hand_pos_x - thumb_x), ...
1e3*(hand_pos_y - thumb_y), ...
1e3*(hand_pos_z - thumb_z)];
x_axis = [1, 0, 0];
y_axis = [0, 1, 0];
z_axis = [0, 0, 1];
% calculate angle between axis
rot_x = (acos((line_v(1,1).*x_axis(1,1))/(norm(line_v)*norm(x_axis))))*(180/pi); %degrees
rot_y = (acos((line_v(1,2).*y_axis(1,2))/(norm(line_v)*norm(y_axis))))*(180/pi); %degrees
rot_z = (acos((line_v(1,3).*z_axis(1,3))/(norm(line_v)*norm(z_axis))))*(180/pi); %degrees
% Plot circle around hand in depth coordinates
if (flagStart == 1)
th = 0:pi/50:2*pi;
rd = round(90 - 1e3*min(hand_pos_z)/30);
coords = depthMetaData.DepthJointIndices(25,:,trackedBodies); % coordinates for depth Map
plot(rd*sin(th) + coords(1,1),rd*cos(th) + coords(1,2),'g'); %plot for depth Map
plot(skeletonJoints(12,1,body), skeletonJoints(12,2,body), 'r*');
end
end
end
% store or display the coordinates of the right hand
hand_x(counter) = double(hand_pos_x(1,1)); %meters
hand_y(counter) = double(hand_pos_y(1,1)); %meters
hand_z(counter) = double(hand_pos_z(1,1)); %meters
f_thumb_x(counter) = double(thumb_x(1,1)); %meters
f_thumb_y(counter) = double(thumb_y(1,1)); %meters
f_thumb_z(counter) = double(thumb_z(1,1)); %meters
if (counter == 1)
distance = [0;0;0]; %distance hand moved (m)
velocity_x(counter) = distance(1,1)/depthTime; %velocity at x axis (m/sec)
velocity_y(counter) = distance(2,1)/depthTime; %velocity at y axis (m/sec)
velocity_z(counter) = distance(3,1)/depthTime; %velocity at z axis (m/sec)
acceleration_x(counter) = velocity_x(counter)/depthTime; %acceleration at x axis (m/sec2)
acceleration_y(counter) = velocity_y(counter)/depthTime; %acceleration at y axis (m/sec2)
acceleration_z(counter) = velocity_z(counter)/depthTime; %acceleration at z axis (m/sec2)
else
sum_time = depthTime - temp_time; %seconds
temp_time = depthTime; %seconds
distance = [hand_x(counter) - hand_x(counter-1);hand_y(counter) - hand_y(counter-1);hand_z(counter) - hand_z(counter-1)]; % distance hand moved at each axis (meters)
velocity_x(counter) = distance(1,1)/sum_time; %velocity at x axis (m/sec)
velocity_y(counter) = distance(2,1)/sum_time; %velocity at y axis (m/sec)
velocity_z(counter) = distance(3,1)/sum_time; %velocity at z axis (m/sec)
acceleration_x(counter) = velocity_x(counter)/sum_time; %acceleration at x axis (m/sec2)
acceleration_y(counter) = velocity_y(counter)/sum_time; %acceleration at y axis (m/sec2)
acceleration_z(counter) = velocity_z(counter)/sum_time; %acceleration at z axis (m/sec2)
end
% Kalman filter estimation (Linear)
% Estimation
if (counter == 1)
Q_estimate = [hand_x(counter);hand_y(counter);hand_z(counter);velocity_x(counter);velocity_y(counter);velocity_z(counter)]; %initial state for x y z Vx Vy Vz
end
Q_loc_meas(:,counter) = [hand_x(counter);hand_y(counter);hand_z(counter)]; %Real location from kinect
u = norm([acceleration_x(counter) acceleration_y(counter) acceleration_z(counter)]); %norm of acceleration of thumb
Q_estimate = A * Q_estimate + B * u; % Predicted next state (linear)
P = A * P * A' + Ex; % Predict next covariance
K = P * C' * inv( C * P * C' + Ez ); % Kalman Gain factor
if ~isnan(Q_loc_meas(:,counter))
Q_estimate = Q_estimate + K * ( Q_loc_meas(:,counter) - C * Q_estimate ); %Update the state estimation
end
P = ( eye(6) - K * C ) * P; %Update covariance
%Store data
Q_loc_estimate = [Q_loc_estimate; Q_estimate(1:3)];
vel_estimate = [vel_estimate;Q_estimate(4:6)];
F_Q_estimate(:,counter) = Q_estimate(1:3);
F_vel_estimate(:,counter) = Q_estimate(4:6);
%Convert from real coordinates to pixel location in depth map to print
depth_u = (depth_focal_length_x * Q_estimate(1) / Q_estimate(3)) + 512/2;
depth_v = -((depth_focal_length_y * Q_estimate(2) / Q_estimate(3)) - 424/2);
%Convert from real coordinates to pixel location in depth map to print
color_u = (color_focal_length_x * Q_estimate(1) / Q_estimate(3)) + 1920/2;
color_v = -((color_focal_length_y * Q_estimate(2) / Q_estimate(3)) - 1080/2);
if (flagStart == 1)
plot(rd*sin(th)+depth_u,rd*cos(th)+depth_v,'r'); % the kalman filter tracking in depth map
end
hold off; % release frame
counter = counter + 1; %counter for storing
time(counter) = time(counter-1) + frame_dt;
flushdata(depthVid); %flush frames to free memory
flushdata(colorVid); % flush frames to free ram
end
if (counter == 70)
break % break after a number of points
end
end
%% Plot graphs for kinect measurements and kalman prediction
% Plot the position graphs
close all
figure('units','normalized','outerposition',[0 0 1 1]);
% X axis
subplot(2,2,1);
plot(time(2:counter),Q_loc_meas(1,:)*1e3,'r');
hold on
plot(time(2:counter),F_Q_estimate(1,:)*1e3,'b');
% Compare to the smooth function
% hold on
% plot(time(2:counter),smooth(Q_loc_meas(1,:)*1e3),'m');
title('X AXIS');
xlabel('Time (sec)');
ylabel('Position at X axis (mm)');
h1 = legend('Kinect Measurement','Kalman Filter Prediction','Theoritical Path');
set(h1, 'Position',[0.6 0.2 0.2 0.2]);
hold off
% Y axis
subplot(2,2,2)
plot(time(2:counter),Q_loc_meas(2,:)*1e3,'r');
hold on
plot(time(2:counter),F_Q_estimate(2,:)*1e3,'b');
% Compare to the smooth function
% hold on
% plot(time(2:counter),smooth(Q_loc_meas(1,:)*1e3),'m');
title('Y AXIS');
xlabel('Time (sec)');
ylabel('Position at Y axis (mm)');
hold off
% Z axis
subplot(2,2,3)
plot(time(2:counter),Q_loc_meas(3,:)*1e3,'r');
hold on
plot(time(2:counter),F_Q_estimate(3,:)*1e3,'b');
% Compare to the smooth function
% hold on
% plot(time(2:counter),smooth(Q_loc_meas(1,:)*1e3),'m');
title('Z AXIS');
xlabel('Time (sec)');
ylabel('Position at Z axis (mm)');
hold off
% Set font size and bold
set(findobj(gcf,'type','axes'),'FontName','Arial','FontSize',12,'FontWeight','Bold', 'LineWidth', 1);
%% Release the kinect
stop(depthVid)
stop(colorVid)