Remigiusz Baran

The efficient real- and non-real-time make and model recognition of cars

Make and Model recognition of cars (MMR) has become an important element of automatic vision based systems. Nowadays, MMR utility is commonly added to traffic monitoring (e.g. Licence Plate Recognition) or law enforcement surveillance systems. Facing the growing significance of Make and Model Recognition of cars we have designed and implemented two different MMR approaches. According to their disparate assumption data of these implementations one is obligated to estimate different car models in milliseconds (with a bit less emphasis placed on its accuracy) while the other is aimed first of all to reach higher classification accuracy. Both the implemented MMR approaches, called Real-Time and Visual Content Classification, respectively, are described in this paper in detail and with reference to other MMR methods presented in the literature. Analyses of their performance with respect to classification accuracy and, in case of the Real-Time approach, to its response time are also presented, discussed and finally concluded.

Quality assessment for a visual and automatic license plate recognition

Video transmission and analysis is often utilized in applications outside of the entertainment sector, and generally speaking this class of video is used to perform specific tasks. Examples of these applications include security and public safety. The Quality of Experience (QoE) concept for video content used for entertainment differs significantly from the QoE of surveillance video used for recognition tasks. This is because, in the latter case, the subjective satisfaction of the user depends on achieving a given functionality. Recognizing the growing importance of video in delivering a range of public safety services, we focused on developing critical quality thresholds in license plate recognition tasks based on videos streamed in constrained networking conditions. Since the number of surveillance cameras is still growing it is obvious that automatic systems will be used to do the tasks. Therefore, the presented research includes also analysis of automatic recognition algorithms.

A Smart Camera for Traffic Surveillance

An intelligent surveillance system based on visual information gathered by smart cameras, aimed at traffic monitoring with emphasis on traffic events caused by cars, is presented in the paper. The system components and their capabilities for automatic detection and recognition of selected parameters of cars, as well as different aspects of system efficiency, are described and discussed in detail. Smart facilities for Make and Model Recognition (MMR), License Plate Recognition (LPR) and Color Recognition (CR), embedded in the system in the form of their individual software implementations, are analyzed and their recognition rates detailed. Finally, a discussion of the system’s efficiency as a whole, with an insight into possible future improvements, is included in the conclusion.

Scalar quantization in the PWL transform spectrum domain

The problem of image compression using the PWL (periodic Walsh piecewise-linear) transform and optimal scalar quantization of the PWL spectrum is presented. A short description of the PWL transform is included. Optimal Lloyd-Max quantization is also presented. The procedure of the presented compression method is based on quantization of a class of scalar variables representing blocks of the image. The effectiveness of the compression method is examined and a visual representation of the results is included.

Signal compression based on zonal selection methods

Compression of signals in the spectral domain consists of the rejection of related transform coefficients and reconstruction of signals by inverse transforms. There exist two basic methods used to select coefficients to be transmitted: threshold selection and zonal selection. In threshold selection the reconstruction is made with a subset of coefficients that are larger than a specified threshold. In the zonal method the reconstruction is made with a subset of coefficients lying in a predetermined zone. Only those coefficients are selected for further processing, the remaining coefficients are set to zero. In this paper, zonal compression techniques performed on chosen signals using different transforms are investigated. The mean square error (MSE) is used as a measure of signal reconstruction quality. Plots of the MSE versus compression percentage are included for comparison of methods.

Face Recognition for Movie Character and Actor Discrimination Based on Similarity Scores

A novel face recognition approach dedicated to discriminate between movie characters and actors is presented in the paper. In presented approach, faces are categorized according to similarities determined with regard to their ORB feature descriptors. Constituent procedures, including face normalization, descriptors computation as well as feature matching and similarity discovering are described accurately. Results of their examination are then also reported in details. An exemplary application of the presented approach within the IMCOP platform is presented as well.

Piecewise-linear subband coding scheme for fast image decomposition

Transform coding methods play a very important role in the development of diverse multimedia technologies. The discrete cosine and wavelet transforms are frequently used in the field of image processing and set up a basis for contemporary image compression standards (JPEG, MPEG). In this paper a novel two-channel piecewise-linear subband coding scheme (PL-SBC) is introduced. Its most important advantage are exceptionally fast and easy for implementation computational algorithms combined with good reconstruction quality. The proposed PL-SBC system is based on the dual filter bank consisting of a single filter at the decomposition stage and a single one at the reconstruction stage. For the presented PL-SBC scheme, the conditions of perfect reconstruction and the piecewise-linear approximation filter bank have been defined. The compression properties of a new subband coding scheme have been analysed for different categories of images and in comparison with well-known signal transforms, such as cosine, piecewise-linear, piecewise-constant and wavelet transforms. The research includes a comparison of the reconstruction error both in objective as well as subjective approach.

The IMCOP System for Data Enrichment and Content Discovery and Delivery

This paper presents an overview of the IMCOP system dedicated to data enrichment and content discovery and delivery. A representative client application -- the DEEP platform -- for which the IMCOP system can serve is also depicted. Foremost categories of IMCOP web services are described in detail and their goals are clarified. A concept of Complex Multimedia Objects derived from the MPEG-7 standard and standing, in parallel, for its extension, is also introduced. Finally, advantages of the IMCOP system, as well as IMCOPs current drawbacks, with an insight into proposals for their future improvement, are discussed.

Public Transport Vehicle Detection Based on Visual Information

Freedom of movement is a major challenge for blind or visually impaired people. Movement in urban environment is for such people a big problem. Hence, the aim of the study presented in this paper is to develop an efficient method for identification of different public transport means. To simplify the solution as well as to avoid the need of infrastructure changes, recognition approaches based on visual information collected by smartphones have been assumed. According to the above, detectors based on Haar-like features as well as selected image processing algorithm have been prepared and analysed. Results obtained during performed tests have been reported and compared. Effectiveness of examined individual approaches has been discussed and concluded. Finally, an insight to possible future improvements has been also included.

A capable multimedia content discovery platform based on visual content analysis and intelligent data enrichment

A new capable content discovery platform based on multimedia data enrichment is presented in this paper. The platform, known as the IMCOP system, refers to the concept of intelligent discovery and delivery of multimedia content. Relevant state-of-the-art solutions are described in detail in the background section. The overall architecture and the main components of the IMCOP system are presented next. An original concept of Complex Multimedia Objects which extend the MPEG-7 standard to hold the processed data and bind it into content related collections is introduced. Selected results of tests illustrating how the IMCOP system performs in terms of responsiveness and stability under a particular workload are reported. Finally, IMCOP’s advantages in reference to other content discovery platforms are discussed and summed up.