The thermo- visual fusion based tracking has been deployed for overcoming the shortcomings of alone vision-based object tracking. The assistance from both domains should be wisely merged so that it should result in a useful practice for object tracking. Several techniques had been developed recently to implement a brilliant fusion, but this undeveloped field still inhibits many unsolved challenges. The proposed method aims at increasing the effectiveness of tracking by bi-modal fusion with the introduction of a new set of rules based upon the parameters generated from the decision of individual modality trackers. This practice helps to achieve output by only a single run of the fusion process in every frame. The method also proposes to use minimal information from individual trackers in normal conditions and incorporates the use of supplementary information from imageries merely in case of diverse working conditions. This procedure, in turn, lessens the computations and hence reduces time to process. The experiments performed on well-known publically available datasets show the advantages of the proposed method over the individual visual domain tracking and other existing states of the art fusion techniques.

In this paper we propose a method for automatic detection of crowd escape behaviour. Motion features are extracted by optical flow using Lucas-Kanade derivative of Gaussian method (LKDoG) followed by robust probabilistic weighted feature pooling operation. Probabilistic feature polling chooses the most descriptive features in the sub-block and summarizes the joint representation of the selected features by Probabilistic Weighted Optical Flow Magnitude Histogram (PWOFMH) and Probabilistic Weighted Optical Flow Direction Histogram (PWOFDH). One class Extreme Learning Machine (OC-ELM) is used to train and test our proposed algorithm. The accuracy of our proposed method is evaluated on UMN, PETS 2009 and AVANUE datasets and correlations with the best in class techniques approves the upsides of our proposed method. 

Human action recognition has a very vast application such as security, patient care, etc. Background cluttering, appearance change due to variation in viewpoint and occlusion are the prominent hurdles that can reduce the recognition rate significantly. Methodologies based on Bag-of-visual-words are very popular because they do not require accurate background subtraction. But the main disadvantage with these methods is that they do not retain the geometrical structural information of the clusters that they form. As a result, they show intra-class mismatching. Furthermore, these methods are very sensitive to noise. Addition of noise in the cluster also results in the misclassification of the action. To overcome these problems we proposed a new approach based on modified Bag-of-visual-word. Proposed methodology retains the geometrical structural information of the cluster based on the calculation of contextual distance among the points of the cluster. Normally contextual distance based on Euclidean measure cannot deal with the noise but in the proposed methodology contextual distance is calculated on the basis of a difference between the contributions of cluster points to maintain its geometrical structure. Later directed graphs of all clusters are formed and these directed graphs are described by the Laplacian. Then the feature vectors representing Laplacian are fed to the Radial Basis Function based Support Vector Machine (RBF-SVM) classifier.

Audio imaging can play a fundamental role in computer vision, in particular in automated surveillance, boosting the accuracy of current systems based on standard optical cameras. We present here a method for object tracking application that fuses visual image with an audio image in the template-matching framework. Firstly, an improved template matching based tracking is presented that takes care of the chaotic movements in the template-matching algorithm. Then a fusion scheme is presented that makes use of deviations in the correlation scores pattern obtained across the individual frame in each imaging domain. The method is compared with various state of art trackers that perform track estimation using only visible imagery. Results highlight a significant improvement in the object tracking by the assistance of audio imaging using the proposed method under severe challenging vision conditions such as occlusions, object shape deformations, the presence of clutters and camouflage, etc.

Recently there has been an increase in the use of thermal-visible conjunction technique in the field of surveillance applications due to complementary advantages of both. An amalgamation of these for tracking requires a reasonable scientific procedure that can efficiently make decisions with sound accuracy and excellent precision. The proposed research presents a unique idea for obtaining a robust track estimate with the thermo-visual fusion in the context of fundamental template matching. This method firstly introduces a haphazard transporting control mechanism for individual modality tracking that avoids unexpected estimates. Then it brings together an efficient computation procedure for providing the weighted output using minimal information from the individual trackers. Experiments performed on publically available datasets mark the usefulness of the proposed idea in the context of accuracy, precision and process time in comparison with the state of art methods.