Professional Documents
Culture Documents
ISSN: 2278-0181
Vol. 2 Issue 1, January- 2013
Abstract Moving object detection is an objects, where a powerful variation method using
important step in the video processing. We propose active contours for computing the optical flow is used.
an implementation of image subtraction algorithm However, the high computational time to extract the
to detect a moving object. Detection of object optical flow and the lack of discrimination of the
motion has been implemented and the velocity is foreground from the background, make this method
been calculated. The algorithm is processed with unsuitable for real time processing.
matlab software and we calculated the distance,
frame per time, velocity. This can be used in speed 2. MOVNG OBJECT DETECTION
measurements in traffic signal monitoring systems.
Moving object detection is useful in various fields like
Keywords- moving object detection, velocity border security forces, traffic signal analysis etc.
estimation, subtraction algorithm There are various algorithms used for detecting
moving objects. The background subtraction detects
moving objects by subtracting estimated background
1. INTRODUCTION
models from images. A moving object detection and
tracking system with a static camera has been
Moving object detection in real time is a challenging
developed to estimate velocity, distance parameter.
task in visual surveillance systems. It often acts as an
For that application we presented subtraction
initial step for further processing such as classification
RRTT
algorithm after performing, noise removal filter, edge
of the detected moving object. In order to perform
detection and calculate centroid, velocity and distance
more sophisticated operations such as classification,
of moving object in scene.
we need to first develop an efficient and accurate
IIJJEE
www.ijert.org 1
International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 1, January- 2013
changes are generated by noise or illumination change cases, distance of the moving object also forms a basis
due to weather condition. Considering the robustness, for it to be considered a background, e.g if in a scene
suitability and reduced data access rate, current one person is close to the camera while there is a
research has focused on the benefit of edges structure person far away in background, in this case the nearby
in processing sequence images. But edges show shape person is considered as foreground while the person
and size variation within frames due to the changes in far away is ignored due to its small size and the lack of
illumination and noise. Moreover, the variations for information that it provides. Identifying moving
different edges are not the same. Without considering objects from a video sequence is a fundamental and
this variation from the environment, detectors output critical task in many computer-vision applications. A
cannot be reliable. To solve this problem, the edge is common approach is to perform background
represented as segment that allows incorporating subtraction, which identifies moving objects from the
knowledge to every edge segment about its motion, portion of video frame that differs from the
shape, position, and size variation. And thus, it helps background.
to model the environment using the statistics of all
edge segments. Thus, in this proposed method, edges Background subtraction is a class of techniques for
are treated differently, i.e., segments having small segmenting out objects of interest in a scene for
motion variation statistic will be matched with a high applications such as surveillance. There are many
threshold, while a low threshold may be used for those challenges in developing a good background
having large motion variation. subtraction algorithm. First, it must be robust against
changes in illumination. Second, it should avoid
Background subtraction is particularly a commonly detecting non-stationary background objects and
used technique for motion segmentation in static shadows cast by moving objects. A good background
scenes. It attempts to detect moving regions by model should also react quickly to changes in
subtracting the current image pixel-by-pixel from a background and adapt itself to accommodate changes
reference background image that is created by occurring in the background such as moving of a
averaging images over time in an initialized period. stationary chair from one place to another. It should
The pixels for which the difference is above the also have a good foreground detection rate and the
threshold are classified as foreground. After creating processing time for background subtraction should be
foreground pixel map, some morphological post real-time.
RRTT
processing operations such as erosion, dilation and
closing are performed to reduce the effects of noise 3.1. Edge Detection
and enhance the detected regions. The reference
IIJJEE
background is updated with new images over time to Edge detection is the first step to recover information
adapt to dynamic scene changes. from images. Edges are the significant local changes
of intensity in an image. Edges typically occur on the
Background subtraction technique often relies on boundary between two different regions in an image.
deriving the probabilistic model of the background. Edge also can be defined as discontinuities in image
When a new image is captured, the difference between intensity from one pixel to another. A typical edge
the image and background model is computed for detector has the following steps: (a) it suppresses noise
moving object detection. Unfortunately, the derivation as much as possible, without destroying the true
of the model is complex and computationally edges; (b) it applies a filter to enhance the quality of
expensive. Although alternative approaches not based the edges in the image, (c) it determines which edge
on probabilistic modeling of the background image pixels should be discarded as noise and which should
exist, they are very specific to applications. Therefore, be retained, (d) it determines the exact location of an
most of existing approaches for moving object edge. An optimal edge detector should satisfy the
detection are computationally heavy and subject to following criteria: (a) the optimal detector must
large delays, adversely affecting the performance of minimize the probability of false positives (detecting
real-time surveillance. spurious edges caused by noise) as well as that of false
negatives (missing real edges), (b) the edges detected
Background subtraction is a computational vision must be as close as possible to the true edges, (c) the
process of extracting foreground objects in a particular detector must return one point only for each true edge
scene. A foreground object can be described as an point; that is, it minimizes the number of local maxima
object of attention, which helps in reducing the around the true edge created by noise.
amount of data to be processed as well as provide
important information to the task under consideration.
Often, the foreground object can be thought of as a 3.2. Canny Edge Detector
coherently moving object in a scene. We must
emphasize the word coherent here because if a person The Canny edge detector is one of the most commonly
is walking in front of moving leaves, the person forms used image processing tool to detect edges from
the foreground object while leaves though having image. It has the following steps:
motion associated with them are considered
background due to its repetitive behavior. In some
www.ijert.org 2
International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 1, January- 2013
The aim is to build a robust and novel moving object The distance travelled by the object is determined by
detection algorithm that can detect object in a variety using the centroid. It is calculated by using the
of challenging real world scenarios. The video is Euclidean distance formula. The variables for this are
organized into frames. Image processing tool box is the pixel positions of the moving object at initial stage
used for obtaining the object by subtracting the to the final stage.
frames. The movement of the object is described using
edge detection techniques. Experiment datasets will be
collected for these scenarios for testing the
performance of the algorithm. where X1 = previous pixel position and X2 = present
pixel position in width Y1 = previous pixel position
1. The input video is converted into frames. We and Y2 = present pixel position in height
capture two consecutive frames, i.e. frame N and
frame N+1. The velocity of moving object is calculated by the
2. We convert each of these frames to gray scale. distance it travelled with respect to the time. Euclidean
3. We subtract frame N from frame N+1, to generate distance formula is used to calculate the distance
the difference image. between the sequences of frames. By using the values
4. We run canny edge detection technique which is of distance with respect to frame rate, the velocity of
preferred over the other techniques like Prewitt and the object is defined.
Sobel image in order to remove noise and to detect the
edges. Velocity = Distance Travelled/Frame Rate
5. We identify moving edge segments and also draw a
bounding box over it.
www.ijert.org 3
International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 1, January- 2013
5. RESULTS AND ANALYIS edge segments from moving edge segments. The
velocity and speed of the moving object can also be
This method was tested on several video sequences. calculated. In our future work, we will incorporate
All images were of size 320240 with background multi image detection and calculating speed for the
motion, illumination change. The proposed system multiple moving objects present in the scene for more
was able to detect almost all of the moving objects in sophisticated vision based applications like airport
the sequences. MATLAB was used as an image security, activity recognition, etc. This work can also
processing environment tool. The video is split into be used along with other detection techniques to detect
frames and then filtering is done in order to remove the kind of the object.
noise. Then canny edge detection technique is applied.
Then subtraction method is applied. The moving REFERENCES
object is identified and is separated from the
background. Every edge segment is matched using a [1]. J. Canny. A computational approach to edge
common threshold. Also every background segment detection. IEEE Trans. Pattern Anal. Mach.
has its own motion statistic. An edge segment that has Intell., 8(6):679698, 1986.
high movement statistic will be matched with a wider [2]. M. A. A. Dewan, M. J. Hossain, and O. Chae.
search area and low threshold. Non-moving edge Background independent moving object
segments will benefit from high threshold and narrow segmentation for video surveillance. IEICE
search region. Transactions, 92-B(2):585598, 2009.
[3]. D. Gutchess, M. Trajkovic, E. Cohen-Solal, D.
Figure 2 shows a sample input frame obtained from Lyons, and A. K. Jain. A background model
the given video and the algorithm is applied and initialization algorithm for video surveillance. In
finally the output is obtained .The output clearly draws in proc. IEEE ICCV 2001, Pt.1, pages 733740,
a bounding box on the moving object present in the 2001.
scene and thus the velocity can be calculated. After [4]. Mahbub Murshed, Adin Ramirez, Oksam Chae.
tracking moving object we calculated time per frame Statistical Background Modeling: An Edge
of video and then with the help of centroid of moving Segment based Moving Object Detection
object which is like a reference point we calculate Approach.in Seventh IEEE International
RRTT
distance of moving object .We also calculate velocity Conference on Advanced Video and Signal Based
of moving object. Surveillance, 2010. pages 300-306, May. 2010.
[5]. J. H. Duncan, and T. C. Chou, Temporal edges:
In matlab we have implemented this image subtraction The detection of motion and the computation of
IIJJEE
algorithm .Whenever in the example, the man is optical flow, in Proc. IEEE 2nd Int. Conf.
moving, it will detect and track movement. Then we Computer Vision, Florida, USA, Dec. 1988, pp.
calculated centroid using matlab command and then 374-382.
found out the distance according to reference point. [6]. S. Denman, V. Chandran, and S. Sridharan,
The detection output is more accurate and thus can Adaptive Optical Flow for Person Track ng, in
significantly improve the performance of video Proc. Digital Image Computing: Techniques and
surveillance based applications. Applications, Cairns, Australia, Dec. 2005, pp.
44-50.
[7]. F. Ranchin, and F. Dibos, Moving objects
segmentation using optical flow estimation, in
Proc. Workshop on Mathematics and Image
Analysis, Paris, France, Sep. 2004.
[8]. L. Li, W. Huang, I.Y.H. Gu, and Q. Tian,
Foreground object detection in changing
background based on color co-occurrence
statistics, in Proc. IEEE 6th Workshop
(a) (b) (c) Applications of Computer Vision, Florida, USA,
Dec. 2002, pp. 269-274.
Figure 2. (a) A sample input image frame [9]. W. k. Pratt, Digital image processing, fourth
(b) canny edge detection is performed Edition, Pixel Soft, Inc.Los Altos, California,
(c) Moving edge segments using our proposed Willy.
method. [10]. R. C. Gonzalez University of Tennessee, Richard
E. Woods MedData Interactive, Digital Image
Processing, Second Edition, Prentice Hall.
6. CONCLUSION AND FUTURE WORK
This paper illustrates the suitability of using this
model to detect moving objects for the video
surveillance based applications. The strength of our
approach lies in the ability to separate background
www.ijert.org 4