If you are a fan of spy films, you have probably come across scenes where intelligence officers attempt to identify or detect an perpetrator using sophisticated image enhancement technology on surveillance camera footage. . While the idea behind surveillance cameras and object detection is the same in real life, unlike movies, there is often a trade-off between the camera’s field of view and its resolution.
Surveillance cameras generally need to have a wide field of view to make detection of a threat more likely. For this reason, omnidirectional cameras allowing a 360 degree capture range have become a popular choice, for the obvious reason that they do not leave any blind spots but also because they are inexpensive to install. However, recent studies on object recognition in omnidirectional cameras show that distant objects captured in these cameras have rather poor resolution, making their identification difficult. While increasing the resolution is an obvious solution, the minimum resolution required, according to one study, is 4K (3840 X 2160 pixels), which translates into huge bit rate requirements and the need for compression. effective image.
In addition, omnidirectional 3D images often cannot be processed in raw form due to lens distortion effects and must first be projected in 2D. “Continuous processing under high computational loads caused by tasks such as detecting moving objects combined with converting 360-degree video to 4K or higher resolutions to 2D images is simply unachievable in in terms of actual performance and installation costs, ”says Dr Chinthaka Premachandra of the Shibaura Institute of Technology (SIT), Japan, who studies image processing.
Addressing this problem in his latest study published in IEEE Sensor Log, Dr Premachandra, along with his colleague Masaya Tamaki at SIT, envisioned a system in which an omnidirectional camera would be used to locate a region of interest while a separate camera would capture its high-resolution image, thus allowing very precise identification of objects. without incurring significant computational costs. As a result, they built a hybrid camera platform consisting of an omnidirectional camera and a pan-tilt (PT) camera with a 180-degree field of view maintained on either side of it. Incidentally, the omnidirectional camera itself included two fisheye lenses sandwiching the camera body, each lens covering a 180-degree capture range.
The researchers used the Raspberry Pi v2.1 camera modules as PT cameras on which they mounted a pan-tilt module and connected the system to a Raspberry Pi 3 model B. Finally, they connected the whole system, the omnidirectional camera, PT cameras and the Raspberry Pi, to a personal computer for overall control.
The operational flow was as follows: the researchers first processed an omnidirectional image to extract a target region, after which its coordinate information was 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 an additional image should be taken.
Researchers mainly performed four types of experiments to demonstrate performance in four different aspects of the camera platform and separate experiments to verify image capture performance for different target object locations.
While they envision that a potential problem might arise when capturing moving objects for which the complementary images might be shifted due to a delay in image acquisition, they also proposed a counter- potential measurement – the introduction of a Kalman filtering technique to predict the future coordinates of the object when capturing images.
“We anticipate that our camera system will create positive impacts on future applications using omnidirectional imaging such as robotics, security systems and surveillance systems,” comments Dr. Premachandra enthusiastically.
Who wouldn’t be excited, when an extra camera can make so much of a difference?
Source of the story:
Material provided by Shibaura Institute of Technology. Note: Content can be changed for style and length.