Loading...
Please wait, while we are loading the content...
Similar Documents
Video Enhancement for ADAS Systems based on FPGA and CNN Platform
| Content Provider | Semantic Scholar |
|---|---|
| Author | Fasih, Alireza Schwarzlmüller, Christopher Kyamakya, Kyandoghere |
| Copyright Year | 2010 |
| Abstract | In road-transportation terminology, ADAS (Advanced Driver Assistance System) is a combination of monitoring-, alarm-, and control systems for increasing safety level in vehicles. Most of ADAS systems are infrastructureindependent and autonomous. This means these systems must work without any external awareness systems on the roads during the drive. Nevertheless, there are some systems which can cooperate with other vehicles (i.e. CACC Cooperative Adaptive Cruise Control) or awareness systems on the road. Camera is the main part of monitoring and observation unit in ADAS systems. In some cases, fusion of camera data (video frames) and LIDAR (Light Detection And Ranging) leads to more safety and confidence in data. Most of ADAS sub systems such as Adaptive Cruise Control (ACC) , Lane Departure Warning (LDW) , Automatic Parking, Collision Warning Avoidance (CWA) systems and Lane Change Assistance are based on video information. Enhancing video frames is possible by common image processing techniques. These classical algorithms with the sequential processing architecture unit take lot of resources and energy. For producers of ADAS, the main challenges for enhancing image quality are speed and manufacturing cost. It is possible to overcome these problems by developing a proper framework and architecture. In this research, we propose a new method of CNN design based on ESL (Electronic System Design). To implement this model we used fixed-point arithmetic technique and discrete model for CNN. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://vi.uni-klu.ac.at/publications/papers/IJSIP2010.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |