Panoramic background modeling for PTZ cameras with competitive learning neural networks

Distributed memory and localized memory in neural networks

A novel approach for pipeline leak fault detection and work status identification based on fuzzy clustering neural network has been studied. This approach do not need construct exact mathematical model. First of all, we preprocess dataset by extended sigmoid function to normalize each input status vector. Together with prior knowledge, a competitive learning neural network is then used to identify work status, and then the structure and detection scheme of the adaptive algorithm were developed to diagnose the leak fault. An experiment was performed at oil pipeline in Shengli Oil Field. We can learn by experiment results that the proposed method has shown the feasibility and effectiveness.

Partial distortion entropy maximization for online data clustering

In real-world environments, the human visual system perceives a wide range of luminance in a scene. However, the full range of tones in a high dynamic range (HDR) scene cannot be displayed on standard display devices due to their low dynamic range (LDR). Therefore, tone mapping is necessary to faithfully display a HDR scene on an LDR display device. Existing tone mapping methods have problems because details and contrast in a scene are not preserved faithfully, and they also distort the colors in a scene. Thus, we propose a tone mapping method for preserving the detail in an HDR scene using a weighted least squares filter, which preserves the global contrast in a scene by using a competitive learning neural network, before applying a tone reproduction operator to preserve the color without shifting the lightness. According to the Helmholtz–Kohlrausch effect, the perception of brightness depends on the lightness, chroma, and hue of a color. For example, the luminance of pixels with specific colors such as red and blue is low in an HDR scene. The proposed method corrects the lightness of pixels according to the color (i.e., lightness, chroma, and hue) of a tone-mapped image. Experimental results with several test sets of images demonstrated that the proposed tone mapping method with contrast preservation and lightness correction is more suitable for dynamic range compression than existing tone mapping methods, while it also preserves the color of a scene in an effective way.

Segmentation of images into textural homogeneous regions is a fundamental problem in an image understanding system. Most region-oriented segmentation approaches suffer from the problem of different thresholds selecting for different images. In this paper an adaptive image segmentation based on vector quantization is presented. It automatically segments images without preset thresholds. The approach contains a feature extraction module and a two-layer hierarchical clustering module, a vector quantizer (VQ) implemented by a competitive learning neural network in the first layer. A near-optimal competitive learning algorithm (NOLA) is employed to train the vector quantizer. NOLA combines the advantages of both Kohonen self- organizing feature map (KSFM) and K-means clustering algorithm. After the VQ is trained, the weights of the network and the number of input vectors clustered by each neuron form a 3- D topological feature map with separable hills aggregated by similar vectors. This overcomes the inability to visualize the geometric properties of data in a high-dimensional space for most other clustering algorithms. The second clustering algorithm operates in the feature map instead of the input set itself. Since the number of units in the feature map is much less than the number of feature vectors in the feature set, it is easy to check all peaks and find the `correct' number of clusters, also a key problem in current clustering techniques. In the experiments, we compare our algorithm with K-means clustering method on a variety of images. The results show that our algorithm achieves better performance.

Surveillance systems are omnipresent in our day-to-day life as these systems ensure our safety, most of the surveillance systems are based on video sensory modality and machine intelligence. Video surveillance systems in real time are very expensive because they require very high processing. Most of the times, these lack reliability and robustness in real-time systems. To confront this issue, we have come up with real-time audio surveillance system. Where the audio signals are picked up using highly sensitive microphones and then the signal is amplified to a great extent. The picked-up signals contain a lot of other noise apart from the desired information. Hence, the procured audio signals are denoised with the help of signal processing technique called spectral subtraction. After denoising, there are chances of the mixing of audio signals so we are applying the speech separation techniques to get separate speech signals from the multi-speaker audio signal, which will help us to get clear insights of what a monitored person is speaking about.

A major challenge that all law-enforcement and intelligence-gathering organizations are facing is to accurately and efficiently analyze the growing volumes of crime data. In many fields, video surveillance can be used for that purpose. Video surveillance systems are now commonly used at various places like banks, hotels, schools, supermarkets. These systems are applicable for real-time monitoring or post checking. Current video surveillance systems have lower intelligence and it need people to monitor them. Today, video surveillance systems are used for traffic cameras, too. Traffic cameras are an innovative and extremely functional application of video surveillance technology. Whether they are recording traffic patterns for the future study and observation or monitoring traffic and issuing tickets for moving violations, traffic cameras are an explosively popular form of video surveillance. One way to find traffic patterns or to prevent traffic accidents and provide better road security is to use data mining techniques. With data mining, stream from many traffic cameras can be processed in real time. Property like this could provide some intelligence upgrade within systems, so they could make decision on their own without the need of people to monitor them. In this paper the authors described the data mining techniques for stream processing, pattern and abnormality detection in traffic surveillance systems. Each video recording and the use of recorded video on the court must be covered by legal regulations. The authors investigated and extracted the appropriate articles from concrete Laws of Republic Serbia, covering this kind of problems.

Video surveillance has demonstrated its value and benefits countless times by providing real-time monitoring of a facility's environment, people, and assets, recording events for subsequent investigation, proof of compliance and audit purposes. This paper evaluates Quality of Service (QoS) in effective on-demand video surveillance system activated from a remote location with cloud-based storage, and provision for predictive analytics, which relies on Tier-4 network integration. The discrete event tool of Riverbed modelling software version 1.75 has been used to evaluate the on-demand real-time video surveillance system based on on-demand latency, throughput and resource utilization using both primary and secondary backups at the storage end. It was observed that the primary backup took about 0.002secs while the secondary backup took about 0.01secs delay, and a throughput of 100packets/sec obtained without virtualization which significantly increased to 900packets/sec (primary) and 1100packets/sec (secondary) over time with virtualization. These values characterize a state-of-the-art high speed network.

The identification of culprits from videosurveillance systems brings about relevant problems for what concerns the reliability of comparison and quantification of the degree of match between the culprit and the suspect. The actual videosurveillance give only 2D images of the culprit, which can be compared to a 3D model of the suspect’s face in 2D-3D superimposition. However, in the next future the videosurveillance systems will provide a 3D image of the filmed persons, which may allow the operator to perform 3D-3D superimposition. This study aims at exposing a pilot study based on 10 male subjects: the face from each individual underwent two acquisitions by stereophotogrammetry. On each 3D facial model, 9 landmarks (right and left: en, ex, ch: midline: se, gn, prn) were identified by VAM® software; the acquisitions of each individual were then superimposed with the models of all the other subjects in order to reduce the distance between the corresponding points. The RMS value (Root Mean Square) between the two models was then calculated. When the two superimposed models belonged to the same individual, RMS value was in mean 1.58 mm (SD: 0.61 mm), whereas when the two models were taken from different persons, the same value was 4.54 mm (SD: 2.02 mm) with a statistically significant (p<0.0001). This pilot study shows that the judgment of identification based on 3D-3D superimposition may provide reliable results. The improvement specific methods of 3D-3D comparison may provide in the next future relevant advantages in this field of forensic anthropology.

Search for criminals becomes more difficult in modern conditions of continuous growth of the population of big urban centers and more and more frequent use of the latest achievements of science and technology by criminals for the realization of their purposes and for counteraction to law enforcement agencies. To increase the efficiency of the search for offenders it is necessary to introduce the latest technologies. Therefore, the systems of video recording and video surveillance, which currently operate in Ukraine and are one of the areas of using these technologies, have been studied. It has been noted that the installed CCTV cameras mostly work in manual mode. The employee of the information processing center must watch the video in order to gather evidence and search for suspects. It takes a lot of time and requires the involvement of a large number of staff, which makes the search for criminals and obtaining evidence from video surveillance systems ineffective.&#x0D; Tendencies in the development of intelligent video surveillance systems in developed countries have been analyzed. It has been found out that such countries (EU, USA, China, Russia) install modern “smart” CCTV cameras, the information from which is sent to modern situational centers, where it is processed by using artificial intelligence, neural networks and cloud infrastructure. Certain types of cameras even have the ability to independently process the received information. This makes it possible to perform biometric facial recognition, to detect behavioral abnormalities of suspects who intend to commit offenses; to record the appearance and disappearance of objects, to recognize the state car plates, to recognize their type, brand and color, to record the way of their movement, to carry out the analysis of these factors on the basis of these factors to search for suspects and cars in the city.&#x0D; Video surveillance centers, in particular within one of the most important areas of public safety – traffic safety – have the ability to recognize prohibited or atypical traffic, to capture the intersection of the restricted area or the intersection of the visual line, to respond to the passage of people in a given direction, to identify traffic conditions, to detect traffic density, congestion and mass congestion of vehicles, to respond to the appearance of people in the surveillance zone, etc.&#x0D; Based on the review of the experience of introduction and use of video surveillance systems in Ukraine and analysis of their development in developed countries, the authors have suggested the ways to modernize and further develop security systems in Ukraine by using CCTV cameras.

Video surveillance systems play a crucial role in ensuring public safety and security. However, the traditional methods of surveillance often fall short in effectively identifying individuals, particularly in crowded or dynamic environments. This research addresses the limitations of conventional video surveillance by proposing a machine learning-based facial recognition system. The increasing demand for robust security measures necessitates the development of advanced technologies in video surveillance. Facial recognition has emerged as a promising solution, but existing systems struggle with accuracy and efficiency. This research aims to bridge these gaps by leveraging machine learning techniques for facial recognition in video surveillance. Conventional video surveillance struggles with accurate and rapid identification of individuals, leading to potential security lapses. This research addresses the challenge of enhancing facial recognition accuracy in real-time video feeds, especially in scenarios with varying lighting conditions and occlusions. While facial recognition has gained traction, there is a significant research gap in the implementation of machine learning algorithms tailored for video surveillance. This study aims to fill this void by proposing a novel methodology that combines deep learning and computer vision techniques for robust facial recognition in dynamic environments. The proposed methodology involves training a deep neural network on a diverse dataset of facial images to enable the model to learn intricate facial features. Additionally, computer vision algorithms will be employed to handle challenges such as occlusions and varying lighting conditions. The model''''s performance will be evaluated using real-world video surveillance data. Preliminary results demonstrate a significant improvement in facial recognition accuracy compared to traditional methods. The machine learning-based system exhibits enhanced performance in challenging scenarios, showcasing its potential for practical implementation in video surveillance systems.

Now a day, video surveillance has gained the most attention in real-time applications. Because of the improvement in domain of machine learning, various methodologies have been designed for multi-object detection. The real-time surveillance systems are the need of various organization as security; hence, this area of research has important. This paper introduces a new methodology for Moving object detection. In the area of real-time video surveillance systems, multi-object detection and object tracking are an important step in video processing. The moving system is critical for many of the application such as operational robot, surveillance systems in military. These real-time surveillance systems are more useful for intelligence gathering, security issues, and many personal needs. Different techniques are used for this task and research is vastly done to make this system automated and to make it reliable. In this research, subjective quality assessment of object detection and object tracking is discussed in detail. In the proposed system, the background subtraction is done from the clean original image and use of CNN with this. We have applied background subtraction method with different filtering algorithms. We have implemented and tried MoD2 and KNN algorithm on the test video and fetched the foreground image. After this, we have applied CNN to track the object. After tracking the objects, we have found out the accuracy with background subtraction method and without background subtraction method. The proposed system improved by approximately 2% as compared to existing techniques.

Everyday events demonstrate (if there is still a need for demo) the need for security. One cause is the occurrence of antisocial events which seem to be once again on the rise. Another cause (increasingly important) is represented by terrorist threats. As events in New York, London and Madrid have demonstrated, this threat is very much alive. Video surveillance is an alternative attempting to prevent the above events to happen (or reduce their chances). This paper presents a digital video surveillance system with distributed IP-cameras, describes the camera HTTP commands to set the parameters and provides and algorithm for automatic camera-configuration.

The subject of study. A new algorithm for real-time data compression in a video surveillance system from an unmanned aerial vehicle is proposed. The algorithm makes it possible to obtain different degrees of video data compression using different codecs for specific video recording conditions (UAV altitude and flight speed, nature of the observed terrain, etc.), and also provides the ability to transmit video data to the flight control point via radio channel. A detailed analysis of the quality of the proposed algorithm for shooting in various video formats that meet international standards (HD and Full HD) was carried out, both in terms of visual perception and objective criteria. The purpose of the study was to conduct a comprehensive research on the problem of improving the quality of the video surveillance system from the UAV and to create a video compression algorithm that allows broadcasting video data to the control center in real time with high quality. Objectives. To develop and programmatically implement a video compression algorithm based on modern codecs for video surveillance sys­tems on board a UAV. To use modern video cameras and a microcomputer platform as hardware. To analyze the quality indicators of the algorithm in different conditions. Methods used. Evaluation of visual perception of video compression results, statistical analysis and processing of the parameters of the analyzed video sequence as a random spatio-temporal process. Results. A new video data compression algorithm for the UAV video surveillance system based on the Raspberry Pi 5 microcomputer using modern video compression codecs H.264 and H.265 was synthesized. The algorithm provides the required compression ratio and data transmission rate over the radio channel between the UAV and the flight control center. Detailed estimates of the quality of the algorithm are obtained. Conclusions. Scientific novelty of the results: A new universal video data compression algorithm for video surveillance systems on board UAVs with the ability to transmit video over a radio channel in real time has been proposed and implemented. The quality of the proposed algorithm is verified on real video surveillance results. The proposed algorithm is implemented in the Python software environment.

Now the visual surveillance system has been studied extensively, and has made great progress, but many surveillance systems are only applied in specific sites, and the level of intelligence is not high. Intelligent video surveillance system mainly carries out research on techniques of detection and tracking of moving targets, which are very important for detection and understanding of abnormal behaviors, so the effect of moving detection directly affects the overall effect of video surveillance systems, moving detection of intelligent visual surveillance is mainly target classification, tracking, detection, behavior recognition and so on. Detection and tracking of moving targets in surveillance applications has greatly improved intelligence, accuracy and reliability of video surveillance and has greatly reduced the burden on staff.