The researchers are constructing a dual hybrid camera system that enables 360-degree surveillance for target acquisition and their acquisition in high resolution
If you’re a fan of spy movies, you’ve likely come across scenes where the intelligence agents try to identify or spot a culprit using sophisticated security camera image enhancement technology. While the idea behind surveillance cameras and object recognition is the same in real life as opposed to filming, there is often a tradeoff between the camera’s field of view and its resolution.
Surveillance cameras usually need to have a wide field of view to make a threat more likely to be detected. For this reason, omnidirectional cameras that provide 360 degree coverage have become a popular choice, for the obvious reason that they don’t leave a blind spot, but also because they are cheap to install. However, recent studies on object recognition in omnidirectional cameras show that distant objects captured in these cameras have a rather poor resolution, making them difficult to identify. While increasing the resolution is an obvious solution, one study found that the minimum resolution required is 4KB (3840 x 2160 pixels), which creates huge bit rate requirements and the need for efficient image compression.
In addition, due to lens distortion effects, omnidirectional 3D images often cannot be processed in raw form and must first be projected onto 2D. “Continuous processing under heavy computational loads caused by tasks such as moving object detection combined with converting 360-degree video with a resolution of 4K or higher to 2D images is slow in terms of actual performance and Installation costs simply cannot be realized ”, says Dr. Chinthaka Premachandra from Shibaura Institute of Technology (SIT), Japan, researches image processing.
Addressing this issue in its latest study published in IEEE Sensors JournalDr. Premachandra, together with his colleague Masaya Tamaki from SIT, considered a system in which an omnidirectional camera would be used to localize an area of interest, while a separate camera would capture the high-resolution image, which would make highly accurate object identification without the possibility of high computational costs . Accordingly, they constructed a hybrid camera platform consisting of an omnidirectional camera and a pan-tilt camera (PT camera) with a 180-degree field of view on both sides. Incidentally, the omnidirectional camera itself comprised two fisheye lenses that enclose the camera body, with each lens covering a 180-degree field of view.
The researchers used Raspberry Pi Camera Modules v2.1 as PT cameras, to which they mounted a pan-tilt module and connected the system to a Raspberry Pi 3 Model B. Eventually, they combined the entire system, the omnidirectional camera, the PT cameras, and the Raspberry Pi, into one personal computer for general control.
The operational sequence was as follows: The researchers first processed an omnidirectional image to extract a target area. Its coordinate information was then converted into angle information (pan and tilt angles) and then transferred to the Raspberry Pi. The Raspberry Pi in turn controlled each PT camera based on this information and determined whether a complementary image should be recorded.
The researchers mainly conducted four types of experiments to demonstrate performance in four different aspects of the camera platform and separate experiments to check image acquisition performance for different target positions.
While they suspect there might be a potential problem with the detection of moving objects where the complementary images could be shifted due to a time delay in the image acquisition, in addition to a possible countermeasure, they have the introduction of a Kalman filter technique to predict the future Coordinates of the object suggested when taking pictures.
“We expect that our camera system will have a positive impact on future applications with omnidirectional imaging such as robotics, security systems and surveillance systems,” says Dr. Premachandra.
Who wouldn’t be excited if an extra camera can do so much?
Title of the original paper: A hybrid camera system for high-resolution resolution of target objects in omnidirectional images
Diary: IEEE Sensors Journal
DOI: https: /
Via the Shibaura Institute of Technology (SIT), Japan
The Shibaura Institute of Technology (SIT) is a private university with locations in Tokyo and Saitama. Since the founding of its predecessor, Tokyo Higher School of Industry and Commerce, in 1927, the company has maintained “learning through practice” as a philosophy in the education of engineers. SIT was the only private science and engineering university selected for the Top Global University Project sponsored by the Ministry of Education, Culture, Sport, Science and Technology and supported by the Ministry for 10 years from the 2014 academic year. The motto, “Promoting Engineers Who Learn From Society And Contribute To Society” reflects the mission to nurture scientists and engineers who can contribute to the sustainable growth of the world by exposing their 8,000+ students to culturally diverse environments in which they are learn to deal with them, work together and collaborate with fellow students from all over the world.
Website: https: /
About Dr. Chinthaka Premachandra from SIT, Japan
Dr. Chinthaka Premachandra is an associate professor at the Institute of Electrical Engineering of the Graduate School of Engineering and Science at Shibaura Institute of Technology in Japan, where he heads the Machine Vision and Robotics Laboratory. He received his PhD from Nagoya University in Japan in 2011. His laboratory researches mainly in the areas of image processing and robotics. He received the FIT Best Paper Award and the FIT Young Researchers Award from IEICE and IPSJ, Japan in 2009 and 2010. To date, he has published more than 130 articles in prestigious journals and conference reports.