Система цифровой обработки телевизионных изображений для мониторинга обнаружения и распознавания статика-пространственных объектов

Аннотация

The relevance and demanding of the dissertation the theme. In the world great attention is paid to creation of the corresponding of digital television images processing system for monitoring in the field of information and communication technologies. Cunent level of scientific and technical progress as one of the important mechanisms of effective safety system puts in the forefront need of target development of the computer view systems such as automatic recognition of images. «Every year in car crashes around 1.3 million people die and 20-50 million people get different kind of injures»1, Methods of digital television images processing and monitoring system are widely used in the developed countries, in particular, in China, the USA, Germany, India. In the Republic Uzbekistan the large-scale attention is given to citizens’ and vehicles’ safety and prevention of different accidents based on scientifically innovation and using modem audio-video technologies. In this matter, essential results are achieved using modernization of vehicles based on radionavigation and television equipment and monitoring the safety, particularly video surveillance system production has started. Although, the development of intellectual detection devices of dynamic and static spatial objects upon the modem demands are considered. Uzbekistan’s Development Strategy considers such tasks as “... maintaining road safety,... implementation of modern information technologies and their use”2. In order to perform these tasks a development of visual monitoring systems, creation of digital television image processing system for monitoring static-spatial objects for vehicles are considered as an important issues. Today one of the important scientific-technical tasks in the world is improvement of digital television images processing systems, as well as systems for automatic detection and recognition of static-spatial objects for vehicles, and considering this circumstance, special attention is paid to target scientific research, in particular identification of the personality according to biometric data, recognition of number plates of vehicles, obstacles detection systems, recognitions of road signs etc. The dissertation is directly serve the implementation of the tasks set out in the following provisions and decrees of the President of the Republic of Uzbekistan: PD-4947 “On Uzbekistan’s Development Strategy” of 7 February 2017, PD-5005 “On measures to radically increase the effectiveness of the activities of the internal affairs bodies, to strengthen their responsibility for ensuring public order, the reliable protection of the rights, freedoms and legitimate interests of citizens” of 10 April 2017 and PR-1442 "On the priorities of industrial development of Uzbekistan in 2011 -2015 years" of 15 December 2010. The aim of study is modeling, development of algorithms and the framework of the process of detection and recognition of static-spatial objects for visual monitoring systems. Scientific novelty of the research as follows: The scale invariant feature transform (SIFT) algorithm of visual monitoring system for static-spatial objects processing and recognition is improved based on an approach of calculation of angles between feature vectors in Euclidian space; Based on the use of advanced Brailion method and calculation of the sum of absolute difference for visual monitoring systems, the algorithm and methods of obstacles detection along the line of vehicles are developed; The static-spatial objects detection and recognition framework based on the modified and advanced scale invariant feature transform algorithm using graphical dataset is developed; The device and the software on detection and recognition of static-spatial objects are developed for visual monitoring systems of spatial around vehicles. Conclusion Doctoral dissertation on the subject of “Digital television image processing system for monitoring the detection and recognition of static-spatial objects” the following conclusions were obtained during the research: 1. A method for the detection of objects on the basis of Braillon’s model using the methods of calculation of the optical flow and the sum of absolute differences, by which obtained high accuracy of detection of 1.5 times compared with the same methods. 2. A universal framework (structure, block diagram) is developed for detection and recognition of spatial objects using appropriate database. For the first time, this universal framework created an opportunity to detect human faces and their emotions, obstacles around the vehicles and creation of the panoramic image, as well as determining the state numbers of vehicles. 3. Device for the detection and recognition of objects for visual monitoring systems around the vehicle. The device is used to detect objects in addition to wide-angle cameras, ultrasonic sonar, which allows 2 times increase the probability of obstacle detection. 4. A database of potential data objects, such as images of human faces, emotions, state registration numbers of vehicles. This reduces the dependency on other databases. 5. Developed software based on the proposed methods to create a panoramic image around the vehicle, detection of obstacles in their way, the recognition of human faces and their emotions, state registration numbers of vehicles. This, in turn, creates opportunity the development of the software market in Uzbekistan. 6. On the basis of the developed methods and software for experimental studies have shown that the use of modified algorithm developed by converting the scaleinvariant SIFT features gives the probability of correct recognition in the allocation of signs on average in the order of 97.5% per cent but the time spent on the recognition of object is an average of 1.4 seconds. 7. Recommendations about use of the methods developed for visual monitoring systems in various spheres of economy of the Republic of Uzbekistan are developed and create in practice a possibility of development of the concept "Safe city" in system of the Ministry of Internal Affairs of the Republic of Uzbekistan. 8. The results obtained on the basis of proper framework developed by the detection and recognition of static-spatial objects for visual monitoring systems have also been introduced and tested in two companies and gave economical effect 60 mln. soums and 20 thousand USD per year.