Gaetano Di Caterina

Smart surveillance system based on stereo matching algorithms with IP and PTZ cameras

In this paper, we describe a system for smart surveillance using stereo images with applications to advanced video surveillance systems. The system utilizes two smart IP cameras to obtain the position and location of objects. In this case, the object target is human face. The position and location of the object are automatically extracted from two IP cameras and subsequently transmitted to an ACTi Pan-Tilt-Zoom (PTZ) camera, which then points and zooms to the exact position in space. This work involves video analytics for estimating the location of the object in a 3D environment and transmitting its positional coordinates to the PTZ camera. The research consists of algorithms development in surveillance system including face detection, block matching, location estimation and implementation with ACTi SDK tool. The final system allows the PTZ camera to track the objects and acquires images in high-resolution quality.

An improved Mean Shift tracker with fast failure recovery strategy after complete occlusion

The effectiveness of the conventional Mean Shift tracking algorithm diminishes for fast moving targets and complete occlusion. In this paper an improved Mean Shift algorithm comprising a fast failure recovery strategy that aims to deal with randomly moving targets and complete occlusion as encountered in crowded scenes is presented. Experimental results show that after complete occlusion or target loss, the new algorithm can effectively recover and continue to successfully track targets in complex scenarios.

Study on Interaction Between Temporal and Spatial Information in Classification of EMG Signals for Myoelectric Prostheses

Advanced forearm prosthetic devices employ classifiers to recognize different electromyography (EMG) signal patterns, in order to identify the user’s intended motion gesture. The classification accuracy is one of the main determinants of real-time controllability of a prosthetic limb and hence the necessity to achieve as high an accuracy as possible. In this paper, we study the effects of the temporal and spatial information provided to the classifier on its off-line performance and analyze their inter-dependencies. EMG data associated with seven practical hand gestures were recorded from partial-hand and trans-radial amputee volunteers as well as able-bodied volunteers. An extensive investigation was conducted to study the effect of analysis window length, window overlap, and the number of electrode channels on the classification accuracy as well as their interactions. Our main discoveries are that the effect of analysis window length on classification accuracy is practically independent of the number of electrodes for all participant groups; window overlap has no direct influence on classifier performance, irrespective of the window length, number of channels, or limb condition; the type of limb deficiency and the existing channel count influence the reduction in classification error achieved by adding more number of channels; partial-hand amputees outperform trans-radial amputees, with classification accuracies of only 11.3% below values achieved by able-bodied volunteers.

Low cost multi-view video system for wireless channel

With the advent in display technology, the 3DTV will provide a new viewing experience without the need of wearing special glasses to watch the 3D scenes. One of the key elements in 3DTV is the multi-view video coding, obtained from a set of synchronized cameras, capture the same scene from different view points. The video streams are synchronized and subsequently used to exploit the redundancy contained among video sources. A multi-view video consists of components for data acquisition, compression, transmission and display. This paper outlines the design and implementation of a multi-view video system for transmission over a wireless channel. Synchronized video sequences acquired from four separate cameras and coded with H.264/AVC. The video data is then transmitted over a simulated Rayleigh channel through Digital Video Broadcasting -Terrestrial (DVB-T) system with Orthogonal Frequency Division Multiplexing (OFDM).

Co-operative surveillance cameras for high quality face acquisition in a real-time door monitoring system

The increasing number of CCTV cameras in use poses a problem of information overloading for end users. Smart technologies are used in video surveillance to automatically analyze and detect events of interest in real-time, through 2D and 3D video processing techniques called video analytics. This paper presents a smart surveillance stereo vision system for real-time intelligent door access monitoring. The system uses two IP cameras in a stereo configuration and a pan-tilt-zoom (PTZ) camera, to obtain real-time localised, high quality images of any triggering events.

Face detection and stereo matching algorithms for smart surveillance system with IP cameras

In this paper, we describe a smart surveillance system to detect human faces in stereo images with applications to advanced video surveillance systems. The system utilizes two smart IP cameras to obtain the position and location of the object that is a human face. The position and location of the object are extracted from two IP cameras and subsequently transmitted to a Pan-Tilt-Zoom (PTZ) camera, which can point to the exact position in space. This work involves video analytics for estimating the location of the object in a 3D environment and transmitting its positional coordinates to the PTZ camera. The research consists of algorithm development in surveillance system including face detection, stereo matching, location estimation and implementation with ACTi PTZ camera. The final system allows the PTZ camera to track the objects and acquires images in high-resolution.

Automatic pharynx and larynx cancer segmentation framework (PLCSF) on contrast enhanced MR images

A novel and effective pharynx and larynx cancer segmentation framework (PLCSF) is presented for automatic base of tongue and larynx cancer segmentation from gadolinium-enhanced T1-weighted magnetic resonance images (MRI). The aim of the proposed PLCSF is to assist clinicians in radiotherapy treatment planning. The initial processing of MRI data in PLCSF includes cropping of region of interest; reduction of artefacts and detection of the throat region for the location prior. Further, modified fuzzy c-means clustering is developed to robustly separate candidate cancer pixels from other tissue types. In addition, region-based level set method is evolved to ensure spatial smoothness for the final segmentation boundary after noise removal using non-linear and morphological filtering. Validation study of PLCSF on 102 axial MRI slices demonstrate mean dice similarity coefficient of 0.79 and mean modified Hausdorff distance of 2.2 mm when compared with manual segmentations. Comparison of PLCSF with other algorithms validates the robustness of the PLCSF. Inter- and intra-variability calculations from manual segmentations suggest that PLCSF can help to reduce the human subjectivity.

Automatic misclassification rejection for LDA classifier using ROC curves

This paper presents a technique to improve the performance of an LDA classifier by determining if the predicted classification output is a misclassification and thereby rejecting it. This is achieved by automatically computing a class specific threshold with the help of ROC curves. If the posterior probability of a prediction is below the threshold, the classification result is discarded. This method of minimizing false positives is beneficial in the control of electromyography (EMG) based upper-limb prosthetic devices. It is hypothesized that a unique EMG pattern is associated with a specific hand gesture. In reality, however, EMG signals are difficult to distinguish, particularly in the case of multiple finger motions, and hence classifiers are trained to recognize a set of individual gestures. However, it is imperative that misclassifications be avoided because they result in unwanted prosthetic arm motions which are detrimental to device controllability. This warrants the need for the proposed technique wherein a misclassified gesture prediction is rejected resulting in no motion of the prosthetic arm. The technique was tested using surface EMG data recorded from thirteen amputees performing seven hand gestures. Results show the number of misclassifications was effectively reduced, particularly in cases with low original classification accuracy.

3-dimensional throat region segmentation from MRI data based on fourier interpolation and 3-dimensional level set methods

A new algorithm for 3D throat region segmentation from magnetic resonance imaging (MRI) is presented. The proposed algorithm initially pre-processes the MRI data to increase the contrast between the throat region and its surrounding tissues and to reduce artifacts. Isotropic 3D volume is reconstructed using the Fourier interpolation. Furthermore, a cube encompassing the throat region is evolved using level set method to form a smooth 3D boundary of the throat region. The results of the proposed algorithm on real and synthetic MRI data are used to validate the robustness and accuracy of the algorithm.

Yawn analysis with mouth occlusion detection

Abstract One of the most common signs of tiredness or fatigue is yawning. Naturally, identification of fatigued individuals would be helped if yawning is detected. Existing techniques for yawn detection are centred on measuring the mouth opening. This approach, however, may fail if the mouth is occluded by the hand, as it is frequently the case. The work presented in this paper focuses on a technique to detect yawning whilst also allowing for cases of occlusion. For measuring the mouth opening, a new technique which applies adaptive colour region is introduced. For detecting yawning whilst the mouth is occluded, local binary pattern (LBP) features are used to also identify facial distortions during yawning. In this research, the Strathclyde Facial Fatigue (SFF) database which contains genuine video footage of fatigued individuals is used for training, testing and evaluation of the system.