Professional Documents
Culture Documents
Image Processing - Using Machine Learning: Software Requirement Specification
Image Processing - Using Machine Learning: Software Requirement Specification
LEARNING
ADITYA CHETTRI
Reg no:- 201600233
KRISHNA KUMAR
Reg no:-
3
2 Overall Descriptions
This section will give an overview of the comparison of object recognition by using
machine learning. This will be explained in its context to show how the component
interacts with other components and introduce the basic functionality of it. It will also
describe how the object recognition will use the system and what functionalities are
available. At last, the constraints and assumptions for the component will be presented.
In our research, we are focusing on a live feed to identifying and detecting unusual
behaviors can be human or objects to prevent threat and notify by the alert as a response.
Kim and Grumman proposed a method to detect abnormalities in a video sequence based
on a space-time Markov random field model. This model dynamically adapts to abnormal
activities that consist of unpredictable variations. Some of the current methods for the
detection of abnormal behavioral patterns are based on unsupervised one-class learning
approaches. Template Matching Evaluating is only a subset of the possible
transformations was considered in the limited context of template matching under 2D
translation [1]. Template matching has various applications and is used in such fields as
face recognition and medical image processing. Systems have been developed and used
in the past to count the number of faces that walk across part of a bridge within a certain
amount of time. Other systems include automated calcified nodule detection within
digital chest X-rays. Recently, this method was implemented in geostatistical simulation
4
which could provide a fast algorithm [2]. Mostly up to now proposed system is just to
seek out the specific shape from a certain image, for an image with several edges, edge
detector rule would possibly take a longer time to make binary image [3]. We will be able
to reduce time when compared to that and to handle noisy images efficiency of the
system will be maintained when compared to previous research.
The automatic detection of abnormal objects can be used to alert security. Abnormal
object means that which can’t be seen inside ATM premises usually. The processes have
to be clearly defined to the system in order to process proactively.
6
2.1.2 Hardware interfaces
The user needs to have a PC with Windows OS is installed in order to install and work
with object detection application. To use the object detection application it requires
minimum 1 GHz RAM.
entire application life cycle, from design to deployment. To implement the application of
abnormal object identification, Visual Studio will be used. It provides all the required
facilities and features to developers as well as a user friendly programming environment.
• MSSQL
7
for building compliance solutions. Conforming to regulatory legislation often impact
show data is stored and how it is accessed. MSSQL will provide all the required facilities
and features to implement the database of abnormal object identification application.
• EmguCV
• draw.io
8
2.1.5 Operations
Windows and abnormal object detection application need to be installed correctly in
to the PC. User & system has to perform following set of actions to interact with
Comparison of knowledge in abnormal object detection component. The basic idea is
listed below:
User should have to perform ATM process without carrying any
abnormal objects, if carrying any abnormal object system have to detect.
In order detect anomaly object first system need to teach/feed the shapes
of abnormal object outline templates to the system.
10s duration video clip -> process it frame wise.
Then system will identify different shapes from image.
And among them system will recognize the abnormal objects.
After understanding the scope and the project solution steps, the challenges and
objectives are identified as:
The first challenge is to recognize the abnormal object through image processing of
individual frames of the real time video from the CCTV camera. The key challenges
of this project is to recognize the abnormal object through the real time video capture
and simulate appropriate functionality. Overcoming various factors like frame rate,
frame size, resolution, background environment, physical characteristics of the
abnormal object. Find the most suitable algorithm to reflect abnormal object
identification and detection actions and coordinates efficiently. Select the appropriate
development methodology that is more suitable for this type of application, to reduce
the risk of errors within the system. Testing the system under three levels namely unit
testing, integration testing and system testing.
System should first connect to internet to retrieve video node from the ATM
premises CCTV.
9
2.2 Product functions
In instances where the template may not provide a direct match, it may be useful
to implement the use of eigenspaces – templates that detail the matching object
under a number of different conditions, such as varying perspectives,
illuminations, color contrasts, or acceptable matching object “poses”. For
example, if the system was looking for an object, the eigenspaces may consist of
images (templates) of objects in different positions to the camera, in different
lighting conditions, or with different expressions [2].
10
Fig.5. Use case diagram - Object
11
Use Case Scenarios
12
Fig.6. Functional architecture
13
derived for background subtraction according to the preference of feature
detection and extraction techniques.
14
2.3 User characteristics
Anomaly object detection is the best solution for (banking industry) ATMs to have a top
of the class security system in place along with video surveillance. ATMs can be
susceptible to criminal activity if it’s not properly protected. Cameras on ATMs should
be effective enough to assist in criminal investigations of ATM transactions. By using the
detection of abnormal object system the banking industry can be protected from the
criminal activity. Important to be proactive, not reactive – to ensure safety of
stakeholders and assets.
2.4 Constraints
The Internet connection is a constraint for this function. Since the input is video to the
system over the Internet, and will use APIs. It is crucial that there is an Internet
connection for the application to function and c# shall be the implementation language.
• We assume that the user has all necessary technologies to perform the object
detection application.
• We assume that the user has the required knowledge on the technology.
2.5.2 Dependencies
• System need to get update with the new versions of the software.
The section 1.5 of this document provides the overview of the proposed system and the
section 2 provides the detailed overall description on the system and requirements. The
section 3 contains detailed requirements that should be followed while design and
implementations.
In the first release of the object detection system is to implement the preliminary and
functional specifications described in the section 3. The methodology of implementing
the system may slightly different than the content described in this document during the
system design however the requirements specified will not be changed and the systems
release will tally with its purpose and objectives.
15
3 Specific requirements
3.1 Classes/Objects
For efficient performance of the system, the system should use machines running
Windows with a minimum speed of 1.80 GHZ.
• RAM
RAM is the memory available for PC to use when running programmers. Amount
of RAM determines how many programmers or applications can be open and
running at the same time. The more RAM the more applications can be used
consecutively without making the machine too sluggish. It is important to get the
best RAM in such case like Image processing or multi-tasking.
16
Requirement of internet connectivity.
Designing of anomaly object identification and detection algorithms.
Reliability is the probability that an application will accurately perform its specified task
under stated environmental conditions. Simply, that is how much a user can depend on
the system. The propose application is developing to provide a reliable and efficient
service to the user. All the latest configurations and technology is used to provide the
service accurate and in less time.
3.4.2 Availability
3.4.3 Security
Security is very important feature for this kind of application because all the details
which are provided to system should be valid and accurate. There is a high security to
overcome threats and keep details in a secure manner.
3.4.4 Maintainability
17
4 Supporting information
4.1 References
[1] S. Hsu, “A Video-Based Abnormal Human Behavior Detection for Psychiatric Patient
Monitoring.”
[7] M. Computing, “Video Surveillance Framework,” vol. 3, no. 5, pp. 673–680, 2014.
18
[9] “https://www.linkedin.com/pulse/cctv-surveillance-human-factors-control-rooms-
oyewole-somorin/.”[Online].Available:https://www.linkedin.com/pulse/cctv-
surveillance-human-factors-control-rooms-oyewole-somorin/.
[11] M. Roads, “Information Bulletin A recommended code of practice for the use of
Closed Circuit Television ( CCTV ),” no. March, p. 8, 2008.
19