HMotion: An Algorithm for Human Motion Detection
1 like514 views
The document presents 'Hmotion,' an algorithm designed for human motion detection using a webcam, which minimizes memory usage by capturing and saving images only when motion is detected. It discusses the technical methods employed, such as frame differencing and threshold filters, to identify motion and highlights the system's limitations, particularly its inability to accurately detect slow-moving objects. The study concludes that the algorithm effectively automates the detection and tracking of human motion, providing a useful tool for applications requiring video monitoring.
HMotion: An Algorithm for Human Motion Detection
- 1. ISSN 22773061 1800 | P a g e A u g u s t 3 0 , 2 0 1 3 HMotion: An Algorithm for Human Motion Detection Roselie B. Alday College of Computer Studies Lyceum of the Philippines University roselie_alday@yahoo.com ABSTRACT Computer memory is expensive and the recording of data captured by a webcam needs memory. In order to minimize the memory usage in recording data from human motion as recorded from the webcam, this algorithm will use motion detection as applied to a process to measure the change in speed or vector of an object in the field of view. This application only works if there is a motion detected and it will automatically save the captured image in its designated folder. Indexing terms/Keywords Algorithm, motion detection, computer application, webcam application Academic Discipline And Sub-Disciplines Computer Science, Information Technology, Systems Development SUBJECT CLASSIFICATION Computer Science TYPE (METHOD/APPROACH) Application Development Council for Innovative Research Peer Review Research Publishing System Journal: INTERNATIONAL JOURNAL OF COMPUTERS & TECHNOLOGY Vol 10, No.7 editor@cirworld.com www.cirworld.com, member.cirworld.com
- 2. ISSN 22773061 1801 | P a g e A u g u s t 3 0 , 2 0 1 3 INTRODUCTION Human body tracking from monocular video sequences is a challenging task. The main task is to retrieve the pose of a person starting from a video taken with a camera (Fossati, 2009). At the heart of our study is the fact that human motions contain significant postures that are easily detected. All of the approaches for motion detection in a continuous video stream are based on comparing of the current video frame with one from the previous frames. One of the most common approaches is to compare the current frame with the previous one which the researches adopted in this study. For this purpose the researchers used “difference and threshold filters”. On this step an image was saved with white pixels on the place where the current frame is different from the previous frame on the specified threshold value. The pixels were counted and if the amount will be greater than the predefined an alarm level can signal about a motion event. Most cameras produce a noisy image so the researchers got motion in such places where there is no motion at all. To remove random noisy pixels, researchers used an Erosion filter. If the object is moving smoothly small changes from frame to frame will be received. When the object is moving so slowly the algorithm will not give any result at all. The most efficient algorithms are based on building the so-called background of the scene and comparing each current frame with the background. There are many approaches to build the scene, but most of them are too complex. The researchers’ approach is to move the background frame to the current frame on the specified amount. Scope and Limitations The study will determine the difference of two pictures. The implementation of the program directed by these challenges were discussed and analyzed. However, every study has its own limitation. This study covers the goal of capturing a motion from webcam and saving it in the designated folder. This study does not cover recording video. Research Literature Zhang (2002) in his research stated that video tracking is a technique that is used to detect and track objects based on some typical features of the objects, and monitor their activities by image sequences taken by video cameras. Video tracking can be used in many areas especially in security-related areas such as airports, embassies, and battlefields. Much research has been done with various hardware environments most based on fixed position still cameras. Motion detection, according to Lee (2008) was a system that uses a more advanced mechanisms stressed that there is no need for the supervisor of a security monitoring system to pay attention to it constantly. In current days, when a security camera is installed, a human being has to be on alert all the time while the camera is on since any little movement can require attention of the supervisor. But most of small objects do not need the supervisor's attention since they could be birds, cats or dogs. In this new system the system only reports the unusual change to the supervisor by calculating the motion and size of objects in the screen. Thus the supervisor can be liberated from the 24-hour concentration duty. Instead he can be only be alerted when the real security threat such as a big moving object like a human intruder appears. Roza in her 2009 study entitled “Motion detection and Real time Tracking” motion detection can be broken down into two main subsystems, motion detection and tracking. Motion detection is the process of determining the movement of an object from two or more successive images. Once the movement detection occurs, calculations are made from two images to determine the type of movement made. The information from the calculations is then used in the tracking of the objects. Tracking involves the prediction of the future position of an object and the movement of the cameras to follow the objects. Motion detection according to Bradshaw (2008) is one of the most useful features of the modern video security control system. With this technology, each image received by the camera is processed by software and compared to the previous image. If the system detects change that exceeds the threshold (a parameter set by the user), the system implements the predefined action. The action could be simply that the video is saved, and/or a cell-phone text message or email is sent, or a relay is activated to turn on lights on the alarm. Fossati (2005) developed human body tracking from monocular sequences. The main task is to retrieve the pose of a person from a video sequence taken with a single camera. Theoretical Framework The researchers following the following ina making a motion detection test using a webcam :
- 3. ISSN 22773061 1802 | P a g e A u g u s t 3 0 , 2 0 1 3 Figure 1 Theoretical Framework Codes for the algorithm //FORM (Code) Private Declare Function capCreateCaptureWindow Lib "avicap32.dll" Alias "capCreateCaptureWindowA" (ByVal lpszWindowName As String, ByVal dwStyle As Long, ByVal X As Long, ByVal Y As Long, ByVal nWidth As Long, ByVal nHeight As Long, ByVal hwndParent As Long, ByVal nID As Long) As Long Private Sub Command1_Click() Picture2.Picture = Picture1.Picture End Sub Private Sub Command2_Click() STARTCAM Load Progress Progress.Show Progress.start Unload Progress Command2.Enabled = False Command3.Enabled = True Picture1.AutoRedraw = True Picture2.AutoRedraw = True Timer1.Enabled = True Timer2.Enabled = True End Sub Private Sub Command3_Click() STOPCAM ProgressBar1.Value = 0 Command3.Enabled = False Command2.Enabled = True Picture1.Picture = LoadPicture("nosignal.bmp") Picture2.Picture = LoadPicture("nosignal.bmp") Label1.Caption = "0%" End Sub Private Sub Command4_Click() Design and development of the program. Set the camera to activate the application program. Detection of human movements using webcam and save to designated folder.
- 4. ISSN 22773061 1803 | P a g e A u g u s t 3 0 , 2 0 1 3 List1.Clear End Sub Private Sub Form_Load() Picture1.Width = 320 * Screen.TwipsPerPixelX Picture1.Height = 240 * Screen.TwipsPerPixelY Picture2.Width = 320 * Screen.TwipsPerPixelX Picture2.Height = 240 * Screen.TwipsPerPixelY Picture1.Picture = LoadPicture("nosignal.bmp") Picture2.Picture = LoadPicture("nosignal.bmp") End Sub Private Function Different(ByVal a As Long, ByVal b As Long) As Boolean 'Checks different of two colors ar = a Mod 256: a = a 256 ag = a Mod 256: a = a 256 ab = a Mod 256: a = a 256 br = b Mod 256: b = b 256 bg = b Mod 256: b = b 256 bb = b Mod 256: b = b 256 sense = 255 - Slider1.Value * 5 Different = (Sqr((ar - br) * (ar - br) + (ag - bg) * (ag - bg) + (ab - bb) * (ab - bb)) > sense) 'formula for counting different End Function Private Sub Form_Unload(Cancel As Integer) STOPCAM SaveSetting "MotionDetect", "Param", "s1", Str(Slider1.Value) SaveSetting "MotionDetect", "Param", "s2", Str(Slider2.Value) End Sub Private Sub Timer1_Timer() 'getting picture from camera SendMessage mCapHwnd, GET_FRAME, 0, 0 SendMessage mCapHwnd, COPY, 0, 0 Picture1.Picture = Clipboard.GetData: Clipboard.Clear stepp = 3 'Grid dense Dim qan, qann As Long qan = 0 qann = 0 For i = 1 To Picture1.Width / Screen.TwipsPerPixelX Step stepp For j = 1 To Picture1.Height / Screen.TwipsPerPixelY Step stepp If Different(Picture1.Point(i * stepp * Screen.TwipsPerPixelX, j * stepp * Screen.TwipsPerPixelY), Picture2.Point(Screen.TwipsPerPixelX * i * stepp, j * stepp * Screen.TwipsPerPixelY)) Then Picture1.Circle (i * stepp * Screen.TwipsPerPixelX, Screen.TwipsPerPixelY * j * stepp), 1, RGB(255, 0, 0) qann = qann + 1 End If Next
- 5. ISSN 22773061 1804 | P a g e A u g u s t 3 0 , 2 0 1 3 Next Label1.Caption = Int(qann * 100 / 910) & "%" 'Counting motion in pracentes ProgressBar1.Value = Int(qann * 100 / 910) End Sub Sub STOPCAM() DoEvents: SendMessage mCapHwnd, DISCONNECT, 0, 0 Timer1.Enabled = False Timer2.Enabled = False End Sub Sub STARTCAM() 'Getting handle of camera window mCapHwnd = capCreateCaptureWindow("WebcamCapture", 0, 0, 0, 320, 240, Me.hwnd, 0) DoEvents SendMessage mCapHwnd, CONNECT, 0, 0 'connecting to camera SendMessage mCapHwnd, WM_CAP_DLG_VIDEOFORMAT, 0, 0 'Calling video format dialog DoEvents Slider1.Value = GetSetting("MotionDetect", "Param", "s1", "0") Slider2.Value = GetSetting("MotionDetect", "Param", "s2", "0") End Sub Private Sub Timer2_Timer() If ProgressBar1.Value > 1 - Slider2.Value * 2 Then Beep SavePicture Picture1.Picture, App.Path + "Detected" + Format(Date, "ddmmyyyy") + "__" + Format(Time, "hhmmss") + ".bmp" List1.AddItem "Saved in " + Str(Time) + " " + Str(ProgressBar1.Value) + "% --> " + Format(Date, "ddmmyyyy") + "__" + Format(Time, "hhmmss") + ".bmp" End If End Sub //PROGRESS CODE Public Sub start() For i = 1 To 15 DoEvents SendMessage mCapHwnd, GET_FRAME, 0, 0 SendMessage mCapHwnd, COPY, 0, 0 Form1.Picture2.Picture = Clipboard.GetData: Clipboard.Clear DoEvents ProgressBar1.Value = ProgressBar1.Value + 1 Next End Sub Private Sub Form_Load() ProgressBar1.Value = 0 End Sub
- 6. ISSN 22773061 1805 | P a g e A u g u s t 3 0 , 2 0 1 3 Results The method developed was used to identify and discriminate the body motion at the same time. If there is a detected motion it will save to the designated folder and if there is no motion it will do nothing. Motion was detected through color sensitivity. It was found out that there is motion if color sensitivity is 50% and above, which was saved upon detection. Findings The researchers found out that motion detection of human using webcam can be automatically detected using color sensitivy. Motion detection works on the basis of frame differencing - meaning comparing how pixels (usually blobs) change location after each frame. There are two ways motion can be detected : 1.) looking for a bulk change in the image, and 2.) motion of the middle mass. The problem with these motion detection methods is that neither detects very slow moving objects, determined by the sensitivity of the threshold. But if the threshold is too sensitive, it will detect things like shadows and save there in designated folder. Motion Detection is applied to a process to measure the change in speed or vector of an object in the field of view. The study entitled “Motion Detection of Human using webcam” This application only works if there is a motion detected and it will automatically save the captured image in its designated folder and if there is no motion captured it will not saved. Conclusions The authors conclude that using a webcam can easily detect the motion of a human. It has been made toward the goal of automatically detecting, identifying and tracking of captured image. The system used the Difference and Threshold filters algorithm to compare and detect the difference of the two pictures. She is a graduate of Bachelor of Science in Statistics at the University of the Philippines in 1986, Master in Computer Science at De la Salle University, Manila in 2000 and is doing her dissertation in Phd in Education major in Mathematics and a member of IACSIT since last year. She had been in the teaching profession in more than 25 years and is the Dean of the College of Computer Studies of Lyceum of the Philippines University in Batangas for 12 years. She had presented research papers locally and internationaly, one of which is Philippine government funded at San Diego California last March 2010. She had visited 9 countries in a period of 11 years during study tours and benchmarking visits.She had been the President of PSITE IV- Philippine Society of IT Education for 3 years and had been the PSITE National Secretary for another 3 years and is currently the National Treasurer this year.