Particle Filtering Based Opticle Flow Computation Model for Crowd Anomaly Detection Using Gaussian Mixture Model
Vijay A. Kotkar1, V. Sucharita2

1Vijay A. Kotkar, Research Scholar, Department of Computer Science & Engineering, K L University, Vijayawada (Andhra Pradesh), India.
2Dr. V. Sucharita, Professor, Department of Computer Science & Engineering, Narayana Engineering College, Gudur (Andhra Pradesh), India.
Manuscript received on 24 March 2019 | Revised Manuscript received on 05 April 2019 | Manuscript Published on 18 April 2019 | PP: 583-591 | Volume-7 Issue-6S March 2019 | Retrieval Number: F03140376S19/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Recently, real-time security applications have attracted researcher, industry and real-time application field due to their unmatched security provisioning tasks and improve the security aspects in private and public places. In this field of security, visual surveillance system plays important role which is generally performed with the help of video-clips and recordings. A surveillance system is considered an application scenario which can perform the early detection of abnormality and threat assessment to protect the human from crowd-related issues and ensure the public safety of humans. Computer vision based image processing applications are widely adopted in these applications where videos can be processed for feature extraction such as people, movement and flow from the video sequence. Several researches have been carried out in this field which is based on the feature extraction and classification process but achieving the desired performance for complex scenarios is still considered as a challenging task. Hence, in this article, we present computer vision based novel technique for crowd-behaviour analysis. the proposed approach is divided into multiple stages where according to the first stage, crowd model is developed using particle filtering scheme and later motion patterns are extracted using SFM (Social Force model) along with a novel clustering scheme which helps to identify the pixel information, later, optical flow and streak line flow computation model is developed, later interest point detection and tracking scheme is applied and distribution of crowd is identified which is later classified using GMM (Gaussians Mixture Model) based approach. The complete experimental study is carried out using MATLAB simulation tool and we present a comparative experimental study which shows that the proposed approach achieves better performance of crowd behaviour detection and classification.
Keywords: Computation Model Opticle Mixture Classification Process Clustering.
Scope of the Article: Open Models and Architectures