High-speed object detection methods
Marta Okoń-Fąfara, Adam M. Kawalec · 2018
Classic methods of detection of objects assume a certain threshold, that separates the echo signal of the object from noise. In case of objects with high linear velocity their echoes may be spread in time and can be under the threshold. The reduction of the threshold should improve a probability of detection, but on the other hand it will increase the probability of false alarm. This side effect can be reduced using special methods usually referred to as Track-before-Detect. Their principle of operation base on the analysis of object’s state estimation in several subsequent scans of sensors’ observation space. Subsequently the estimates of the object position are combined in accordance to assumed strategy. If the parameter of association cost achieves the threshold, it means that an object is detected. There are many TBD strategies described in the literature, but two of the most popular are the Multiple Hypothesis Tracking and the Dynamic Programming. The paper presents their principles of operation and a comparison of their effectiveness.