Time of Flight
Fraunhofer Institute for Integrated Circuits
Embedded multi-sensor camera system for ToF measurement and 3D data analysis
- Figure 1. ToF concept
The purpose of this project is a design and implementation of a multi-sensor camera system combining HD and Time-of-Flight (ToF) sensors.
The principle of Time of Flight consists in measurement of the phase-shift between the emitted modulated light impulse and its reflection from detected objects captured by the sensor. On this basis the distance from the object is calculated per pixel, which results in 3D representation of detected scene. The concept is presented in Figure 1.
The ToF operation, by its nature, results in some problems that need to be taken into consideration, such as ambiguity of measurements at larger distances resulting in restricted range of operation, or some difficulties in capturing of fast moving objects.
Commercially available ToF cameras have no or little built-in intelligence and therefore need to be connected to a PC or other computational system. All the required calculations need to be performed by an external computer. The purpose of our project is to build an independent embedded camera system, powerful enough to perform the calculations on the platform without the necessity of an external computational system.
Based on our preliminary functional model (see Figure 2), the multisensor platform will have the following characteristics:
- Capable of multi-sensor functionality due to a modular frontend (HD and ToF)
- Highly computationally powerful embedded platform, incorporating multi-core DSP and a high-performance 7th series Xilinx FPGA
- Universal platform for embedded image processing
- Equipped with additional sensors such as accelerometer to compensate on ToF measurements’ difficulties of moving camera
- Compact design suitable for mobile applications
- Figure 3. Person detection without and with distance information
One of the target applications of the project is video-surveillance. The main aim is to provide more reliable detection of objects than in the case of traditional 2D object recognition systems by making a use of distance information. This can be particularly useful to detect objects which are partially hidden by other objects (Figure 3.). ToF cameras are also able to generate more accurate background models than systems based on other principles such as stereoscopic cameras.
The embedded camera system is designed to cover a wide range of applications, which include:
- Medicine technology: 3D-endoscopy, patient positioning
- Highspeed-TOF: measurements of mechanical vibrations
- Underwater studies: measurements of shipwrecks or geological formations, detection of individual fish or dolphins based on their surface structure, measurements of heavy see, wave heights, etc. (it should be noted that IR light is absorbed by water, therefore other type of illumination is considered)
Set Bookmark