| Summary: | 碩士 === 國立彰化師範大學 === 電機工程學系 === 98 === In a radar tracking system, the multiple-target tracking (MTT) is an indispensable crucial technique. Whether the tracking system is able to correctly estimate the true target trajectory which involves two key problems that the data association technique and maneuvering detection. In this thesis, proposed applying special operation structure of competitive neural network to develop a computing processing for data association technique, to complement the radar tracking system.
This thesis also applies image processing techniques to consider the type and shape of targets, which will increase the detection probability of targets and reduce the estimation error. After pre-processing for image, extraction the features of targets and then use similarity measurement functions for image features recognition. Finally, we can obtain the targets’ image quantity information by Structural Similarity (SSIM), to combine with CHNN data association technique and adaptive processing. This approach can track the target more accurately. The simulation results indicated that the proposed approach can reduce the tracking error and solve the data association problems under complex conditions.
|
|---|
