Video monitoring systems basically collect and record video content recorded by surveillance cameras. When combined with image analysis technology, however, these systems can be transformed into tools that improve the efficiency and functionality of safety and security services. NEC is developing metadata analysis technology for video monitoring systems that can be utilized for IT safety/security purposes. This technology is capable of efficiently extracting desired information from large amounts of video content collected by surveillance cameras.
Super Resolution technologies enable fine magnification of surveillance camera images for purposes such as face and license plate recognition.
Conventional technology requires numerous extracted still images to improve the resolution of subjects in video content and enable clear subject magnification. When magnified by more than 2 or 3 times (4 to 9 times the number of pixels), however, these images become blurred. Therefore, there has been significant demand for technologies that could further improve resolution and enable greater image clarity at higher magnification.
NEC's new technology creates a super resolution image from a single extracted image (of a person's face, license plate, etc.) by utilizing a database (library) of categorized images. These images maintain fine details even when magnified by more than 4 times (more than 16 times the number of pixels), making it possible to distinguish small and distant subjects much more easily than with conventional technologies. NEC’s new technologies can be efficiently teamed with surveillance cameras to cover such large areas as airports and traffic intersections.
Creates images with fine details when highly magnified
These technologies utilize images of subjects from a large library of images stored at various resolutions to create super-resolution images. The best images, at the most appropriate resolution, are automatically selected for use.
Creates customized image libraries
These technologies can efficiently extract small, optimized image libraries from huge libraries of images for specific purposes. Redundant images are eliminated to make the library as small and efficient as possible without compromising image quality.
NEC aims to expand the use of these technologies into a broad range of fields, including the enhancement of satellite and medical images, while actively promoting the development of technologies that contribute to the safety and security of daily life.
Blur and noise caused by motion blur and defocusing are automatically detected and removed.
Luminance information is analyzed and real-time contrast enhancement (image compensation processing) is performed to effectively extract target objects even from dark images of moving subjects that are otherwise difficult to confirm visually.
Moving Object Retrieval
This technology helps investigators search for specific persons and/or vehicles in archived video content recorded by surveillance cameras, etc.
Through years of research, NEC developed proprietary Learning Algorithms and high-performance pattern recognition techniques that can accurately identify extracted moving objects.
World-class Pattern Recognition Method
Automatically extract people, cars (standard, large, bus) and bikes from video
Automatically classify extracted cars as standard, large or bus
Acquire direction data from people, cars and bikes appearing in video
Automatically determine the colors of cars and clothes
Acquire the time and date (Timestamp) of extracted data
Subject Extraction via Background Sequential Learning and Action Prediction Technology