Faced with the proliferation of tiny cameras hidden in accommodation places, and in particular those in South East Asia, security researchers are devising new methods to detect these unwanted cookies.
A year ago, Baidu experts, for example, created an Android application that can detect their Wi-Fi traffic. But such a device does not work for cameras that record their feed locally. Moreover, it does not allow them to be located in a room.
This is why security researchers from the universities of Singapore and Yonsei have just proposed a new, more general method. Called “LAPD” (Laser-Assisted Photography Detection), it is based on ToF sensors that smartphones have been integrating for several years.
Thanks to the emission-reception of infrared laser beams, they make it possible to calculate the distance between the camera and objects and, for example, to calculate a nice bokeh effect.
But when these infrared rays hit a camera lens, a side effect called “lens-sensor retro reflection” (LSRR) can occur. All of the radiation is then reflected and will generate signal saturation at the ToF sensor.
Based on this observation, the researchers created an application that can scan an area to detect the possible presence of a camera lens.
How it works ?
The software will first help the user to position himself well in relation to the objects to be analyzed, because the LSRR effect is only manifested if we are sufficiently close and if the angle of reflection of the infrared rays does not exceed 20 °.
Then the user will have to slightly shift his device several times in a row, so that the application can take different measures. These are then processed by filters and a convolutional neural network, in order to eliminate false positives.
The researchers carried out a whole series of tests, camouflaging various miniature cameras in various objects. They obtain a detection rate of 88.9% for the LAPD app, against 62% for a dedicated infrared ray detection device and 46% for detection with the naked eye.
Asked by The Register, the researchers said they would soon release the source code for their application.
Source : Study report