An Efficient Approach on Constant Human Motion Detection in Survileicne

DOI : 10.17577/IJERTCONV6IS08028

Download Full-Text PDF Cite this Publication

Text Only Version

An Efficient Approach on Constant Human Motion Detection in Survileicne

R. Vanitha

Department of Elctronicsand Communication Engineering, Care Group of Institutions,

Trichy, Tamilnadu, India

Abstract– Video handling is a standout amongst the most difficult zones in picture preparing. It manages distinguishing a protest of intrigue. Movement identification has been utilized as a part of numerous fields either straightforwardly or by implication. In this paper a proficient way to deal with continuous movement recognition in video observation utilizing shading highlight extraction administrator. Utilizing this approach, we enhance the foundation subtraction and recognizing the moving item with more noteworthy precision. In this paper, foundation demonstrating is done keeping in mind the end goal to make the refresh of foundation because of light brightening and change in the climate condition. Forefront identification is done before refreshing the foundation demonstrate. Shading highlight extraction is done with a specific end goal to keep away from the dynamic foundation, for example, moving leaves, rain, snow, undulating water

Keywords: Motion detection, background subtraction, background modeling, dynamic background


    Moving object detection is the basic step for further analysis of video. It handles segmentation of moving objects from stationary background objects. Commonly used techniques for motion detection are background subtraction, statistical methods, temporal differencing and optical flow. Due to dynamic environmental conditions such as illumination changes, shadows and waving tree branches in the wind, object segmentation is a difficult and significant problem that needs to be handled well for a robust visual surveillance system.

    Object classification step categorizes detected moving objects into predefined classes such as human, vehicle, clutter, etc. It is necessary to distinguish objects from each other in order to track and analyze their actions reliably. There are two major approaches toward moving object classification: shape-based and motion-based methods. Shape- based methods make use of objects 2D spatial information whereas motion-based methods use temporal tracked features of objects for the classification solution. The next step in the video analysis is tracking, which can simply be defines as the creation of temporal correspondence among detected objects from frame to frame. This procedure provides temporal identification of the segmented regions and generates cohesive information about the objects in the monitored area. The output produced by the tracking step is generally used to support and enhance motion segmentation, object classification and higher level activity analysis. The final step of the fully automatic video

    surveillance systems is to recognize the behaviors of objects and create high-level semantic descriptions of their actions. The outputs of these algorithms can be used both for providing the human operator with high level data to help him make the decisions more accurately and in a shorter time, and also for offline indexing and searching stored video data effectively.

    In this paper, we aimed to design an efficient algorithm to extract moving objects in survileince. The key of background subtraction is to build and maintain a background model to represent the background of a video, which is a challenging task owing to that backgrounds of scenes in real-life are usually dynamic, including noise, illumination changes, swaying trees, rippling water and so on


    Feature based detection is based on identifying the points of interest in an image such as edges, corners, color compositions, blobs, their points (corners) and ridges. Feature based methods are generally implemented on individual images rather than a sequence of images. The core algorithm in these methods being divided into two categories, 1) extract features 2) classify these features and trains a system for recognition and classification. Feature (specific structures such as points, edges, curves, boundaries etc.) selection is very important as the rest of the algorithm depends on how good the features are detected [4]. Background subtraction is particularly a commonly used technique for motion segmentation in static scenes [2]. It attempts to detect moving regions by subtracting the current image pixel-by-pixel from a reference background image that is created by averaging images over time in an initialized period. The pixels where the difference is above a threshold are classified as foreground. After creating foreground pixel map, some morphological post processing operations such as erosion, dilation and closing are performed to reduce the effects of noise and enhance the detected regions. The reference background is updated with new images over time to adapt to dynamic scene changes.

    There are different approaches to this basic scheme of background subtraction in terms of foreground region detection, background maintenance and post processing.In [3], Heikkila and Silven used the simple version of this scheme where a pixel location (x, y) in the current image It is marked as foreground if the inequality is satisfied,

    Where T is a pre-defined threshold. The background image Bt is` updated by the use of a first order recursive filter Bt+1=It+(1-)Bt where is an adaptation coefficient. The


      basic idea is to integrate the new incoming information into

      the current background image. The larger it is, the faster new changes in the scene are updated to the background frame. However, cannot be too large because it may cause artificial tails to be formed behind the moving objects. The foreground pixel map creation is followed by morphological closing and the elimination of small-sized regions. Although background subtraction techniques perform well at extracting most of the relevant pixels of

      Background Initialization

      Background Maintenance

      Background Model

      Current Frame

      Foreground Detection

      Foreground Mask

      moving regions, they are usually sensitive to dynamic changes when, for instance, stationary objects uncover the background (e.g. a parked car moves out of the parking lot) or sudden illumination changes occur.


In proposed system, the video arrangement initially changed over into outlines as a preprocessing strategy. In customary path there will be need of standard foundation as a source of perspective edge. With this approach, it is conceivable to distinguish new questions in the scene regardless of whether they all of a sudden quit moving. It is likewise conceivable to identify objects that have expelled from the scene. Be that as it may, the settled reference foundation might be not material to the scene alongside the enlightenment variety. In this manner, the precise foundation picture and a high caliber and light resistance foundation refreshing instrument winds up vital for moving article recognition. After that refresh the background for each subtraction made for the frames.


Block diagram 2- background modeling


    Histograms are computed based on the texture over surrounding regions, though that each pixel is modeled identically, its still block-wise. On one hand, its robust to dynamic background such as waving trees and rippling water; on the other hand, it has common drawbacks of block-wise models. A major problem is that the contour of detected object is illegible Because of using histogram over regions, not only the real foreground, but also the background pixels near the edges of foreground will be classified into foreground, and thus the contour of foreground objects is obscured. To reduce the false detection, pixel wise masking i is applied to the output of the background modeling. According to the above modeling, color and intensity of each Background Modeling

    Background model Initialization



    Background Model




    Fig.3 Histogram oriented edge detection for background

    Background Model update


    Adaptive threshold

    Foreground Background Classification

    Illumination Effect


    Foreground Background Classification


    The modified moving average (MMA) is used to compute the average of frames 1 through K for the initial background model generation. For each pixel (x, y), the corresponding value of the current background model Bt(x, y) is calculated using the formula as follows:

    Bt(x,y)=Bt1(x,y)+1/t(It(x,y) Bt1(x, y))

    Where Bt1(x, y) is the previous background model, It(x, y) is

    Foreground Mask

    Block diagram 1- forground and background detection

    the current incoming video frame, t is the frame number in

    the video sequence, and K is experimentally set at 50 to represent the initial background model. In order to reduce frame storage consumption [46], the initial background model adopts the calculated average. This is accomplished by making appropriate use of MMA which holds only the last background model Bt(x, y) and the current incoming video frame It(x, y) during the calculation procedure


    This procedure is used to quickly find a great quantity of background candidates by determining whether or not their respective pixel values for the incoming video frame It(x, y) are equal to the corresponding pixel values of the previous video frame It1(x, y). If the values correspond, it indicates good candidate selection.

    Fig 3a Instant video frame Fig.3b.Previous video frame

    All pixels from the set of background candidates selected via the rapid matching procedure. The pixel with most recent set will be expressed as follows:

    Mt(x, y) = Mt(x, y) + p,

    if It(x, y)>Mt1(x, y)

    Mt(x, y) p, if It(x, y) <Mt1(x, y)

    Where Mt(x, y) is the corresponding pixel within the most recent set of background candidates, Mt1(x, y) is the corresponding pixel within the previous set of background candidates, and p represents the real value which is experimentally set at 1. Notice that the initial background candidate value M0(x, y) is set at I0(x, y).


    Each optimum background pixel of Mt(x, y) will then be supplied to every frame of the background model Bt(x, y). Based on BM, the best possible background pixels are then updated for the background model. Here, we adopt a simple moving average method in order to smooth the proposed background model. This will yield better results when motion detection is performed. The moving average formula is expressed as follows:

    Bt(x,y)=Bt1(x,y)+1/(It(x,y) Bt1(x, y))

    where is the predefined parameter and, in this paper, is experimentally set at 8.


    Fig.6a detection of foreground pixel Fig.6b.bounded value foreground motion detection.

    The background scene related parts of the system is isolated and its coupling with other modules is kept minimum to let the whole detection system to work flexibly with any one of

    the background models. Next step in the detection method is detecting the foreground pixels by using the background model and the current image from video. This pixel-level detection process is dependent on the background model in use and it is used to update the background model to adapt to scene changes. Also, due to camera noise or environmental effects the detected foreground pixel map contains noise.


    After detecting foreground regions and applying post- processing operations to remove noise and shadow regions, the filtered foreground pixels are grouped into connected regions (blobs) and labeled by using connected component labeling algorithm [5]. After finding individual blobs that correspond to objects, the bounding boxes of these regions are calculated. Figure.:7 shows sample foreground regions after region connecting and labeling. Different connected components are shown in different colors in the right image


    Once we have segmented regions we extract features of the corresponding objects from the current image. These features are size (S), center-of-mass or just centroid (Cm) and color histogram (Hc). Calculating the size of the object is trivial and we just count the number of foreground pixels that are contained in the bounding box of the object. In order to calculate the center-of-mass point, Cm = (xCm, yCm), of an object O, we use the following equation:

    Where n is the number of pixels in O. The color histogram, Hc, is calculated over RGB intensity values of object pixels in current image. In order to reduce computational complexity of operations that use Hc, the color values are quantized. Let N be the number of bins in the histogram, then every bin covers 255/N color values. The color histogram is calculated by iterating over pixels of O and incrementing the stored value of the corresponding color bin in the histogram, Hc. So for an object O the color histogram is updated as follows:

    Finally, the histogram obtained would be a 3D matrix (N × N × N) where N, as defined before, is the number of bins. One more blob feature, which would come to use for us in the classification step is compactness(C) [4] which is defined below in equation


    This feature represents how stretched out a shape is. The perimeter is the number of pixels that are part of an object and have at least one 4-connected neighbor that is not in the object. For example, circle has the minimum perimeter for a given area, hence exhibits highest compactness. shows some features, for instance, which were obtained from the final enhanced foreground pixel map

    Fig 7.motion detection using center point IV.CONCLUSION

    In this paper, we aimed at subtracting background and detecting moving objects from video. A whole-body object level tracking system which successfully tracks objects in consecutive frames. The approach we followed is a deterministic one, and it makes use of features like size, color feature, edge feature and histogram oriented gradient.


    1. C. Faloutsos, R. Barber, M. Flickner, J. Hafner, W. Niblack, D. Petkovic, and W. Equitz, Efficient and Effective Querying by Image Content" Journal of Intelligent Information Systems Vol 3, Issue 3-4 ,1994.

    2. Janne Heikkilä, Olli Silvén, "A Real-Time System for Monitoring of Cyclists and Pedestrians," vs, pp.74, Second IEEE Workshop on Visual Surveillance, 1999.

    3. M. McIvor. "Background subtraction techniques". In Proc. of Image and Vision Computing, Auckland, New Zealand, 2000.

    4. Furat N. TawfeeqReal Time Motion Detection in Surveillance Camera Using MATLAB International Journal of Advanced Research in Computer Science and Software Engineering Volume 3, Issue 9, September 2013

    5. Friedman and S. Russell, Image segmentation in video sequences: A probabilistic approach," in Proceedings of the Thirteenth Annual 161Conference on Uncertainty in Artificial Intelligence (UAI-97), pp. 175 to 181, Morgan Kaufmann Publishers, Inc., (San Francisco, CA), 1997.

    6. C. Gonzalez and R. E. Woods. Digital Image Processing. Pearson Education, 2002

    7. H. Kakiuch, et al., Detection Methods Improving Reliability of Automatic Human Tracking System, 2010 4th InternationalConference on Emerging Security Information, Systems and Technologies,Washington DC, 2010, pp. 240-246.

    8. S. Kwak and H. Byun, Detection of Deominant Flow and Abnormal Events in Surveillance Video, Optical Engineering, Vol. 50, No. 2, 2011. pp. 1-8.

    9. Martines Martin E., and del Pobil P, Robust Motion Detection in Real-Life Scenarios (Springer, 2012).

    10. R. Shaalini& C. Shanmugam, R. Niranjan Kumar,, Human Motion Detection and Tracking for Real- Time Security SystemVol.3, Issue 12, December 2013, Shih Chia Huang, An Advanced Motion Detection Algorithm with Video Quality Analysis for Video Surveillance Systems IEEE Transaction on Circuits and Systems for Video Technology, Vol. 21, January. 2011

    11. Aji Joy, Amity P Somaraj, Amal Joe, Muhammed Shafi, Nidheesh TM, Ball Tracking Robot Using Image Processing and Range Detection,IJIRCCE, Vol. 2, Issue 3, March 2014, ISSN: 2320- 9801.

    12. S. Qiu, G. Wen, and Y. Fan, Occluded Object Detection in HighResolution Remote Sensing Images Using Partial Configuration Object Model, in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 10, no. 5, pp. 1909-1925, May 2017

    13. H. Shin, J. Gwak, J. Yu and M. Jeon, Feature flow-based abnormal event detection using a scene-adaptive cuboid determination method, 2016 International Conference on Control, Automation and Information Sciences (ICCAIS), Ansan, 2016, pp. 205-209.

    14. B. Andò, S. Baglio, C. O. Lombardo and V. Marletta, An Event Polarized Paradigm for ADL Detection in AAL Context,in IEEE Transactions on Instrumentation and Measurement,vol. 64, no. 7, pp. 1814-1825, July 2015

    15. X. Cao, L. Yang and X. Guo, Total Variation Regularized RSPCA for Irregularly Moving Object Detection Under Dynamic Background, in IEEE Transactions on Cybernetics, vol. 46, no. 4, pp. 1014-1027, April 2016.

    16. R. C. Luo and C. C. Lai, Multisensor Fusion-Based Concurrent Environment Mapping and Moving Object Detection for Intelligent Service Robotics, in IEEE Transactions on Industrial Electronics, vol. 61, no. 8, pp. 4043-4051, Aug. 2014.

    17. M. Bertini, A. Del Bimbo, and L. Seidenari, Multi-scale and real time non-parametric approach for anomaly detection and localization, in CVIU, 2012

    18. C. Chen, Y. Shao and X. Bi, Detection of Anomalous Crowd Behavior Based on the Acceleration Feature, in IEEE Sensors Journal, vol. 15, no. 12, pp. 7252-7261, Dec. 2015.

    19. Pushpa, D., H.S. Sheshadri, Homography based Multipl Moving Objects Using Multi-Dimensional Event Feeds, International Journal of Computer Science and Information Technology Security, Vol. 3 No.5, October 2013

    20. T. Wang and H. Snoussi, Detection of Abnormal Visual Events via Global Optical Flow Orientation Histogram, in IEEE Transactions on Information Forensics and Security, vol. 9, no. 6, pp. 988-998, June 2014

    21. J.K.Aggarwal, andM.S.Ryoo, Human activity analysis: a review, Journal ACM computing surveys, Vol. 43, 2011.

    22. R. Arroyo, J. J. Yebes, L. M. Bergasa, I. G. Daza, and J. Almazanssssss,Expert video-surveillance system for real-time detection of suspicious behaviors in shopping malls, Expert systems with Applications, Vol. pp 7991-8005, 2015.

    23. B.M.A Amer, and S.A.R Al-Attas, Smart surveillance using

PDA Word Academy of Science, Engineering Technology,Vol.66, pp.251255, 2010.

Leave a Reply