Sign Detection System for Real Time Applications
There are two major parts to detecting an object in an image. The first step is to extract all the important information out of the image and leave out all the noise. This process is called feature extraction. The second part is using these features to detect whether the object is in the image. Both of these part take a large amount of time, especially as the image resolution get bigger, so there are many different techniques to reducing the time these algorithms take.
For our senior thesis we to created a system that would reduce the time it takes to extract the features from an image for the purpose of detecting if a stop sign was in the image. Image processing can often be very slow for normal computers because modern images are often very big. To speed up the detection rate of our algorithm we chose to use a System on a Chip (SoC) device. This device allows us to implement some parts of the algorithm on a Field Programmable Gate Array (FPGA) and other parts on a Hard Processor System (HPS). Because both of these systems are on the same chip, there is a very fast communication speed between the two systems. Using both both an FPGA and a HPS we can significantly speed up the feature extraction process.
This harmony between hardware and software will allow us to extract the feature of an image in real time. Using this we can use a detection algorithm to detect stop signs in video. This means that as the camera is recording the algorithm can determine if a frame has a stop sign in it before the next frame is captured. Real time feature extraction system such as this are very important in high speed applications such as semi-autonomous and driver assisted cars. These cars need to be able to detect objects in their path at high speeds, and very high accuracy rates.