Font Size: a A A

The Study On The Application Of Target Recognition For Low Resolution Radar

Posted on:2012-03-26Degree:MasterType:Thesis
Country:ChinaCandidate:H M MiFull Text:PDF
GTID:2178330335954425Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Radar Target Recognition is an important research direction. It is the technology which uses radar to detect single or multiple targets, and then analyses information obtained, to reveal the electromagnetic scattering mechanism and determine the types and other attributes of the targets. The active low resolution radar is not with the radial and the horizontal high-resolution, so it is difficult to reveal detailed information and to achieve recognition of the target, and it is a difficulty in the area of Radar Target Recognition. However, it is feasible to accomplish rough recognition, making full use of the limited information carried on the echo. There are already some fruitful results and useful methods at home and aboard. Therefore, the study on the application of target recognition for low resolution radar has practical significance.There are two methods to complete a template database. One is field measurement, which has limitations, and the other is computer simulation. Simulation technology can be used as a supplementary means to construct a template database. In this paper, we describe the simulation model of radar target echo signal and sea clutter signal. Similarity coefficient and Mahalanobis distance are used to conduct a validation of simulation echo.In this paper, we study how to extract effective features to characterize the radar target echo. Firstly, we extract the feature vector, including echo wide, shoulder wide, echo energy, and etc., based on the echo shape, which describe directly the target structure along the line of radar sight. Secondly, we achieve the feature extraction based on wavelet transform. The statistical moment features based on discrete wavelet transform matrix, wavelet energy features of each band and the energy fluctuation characteristics are extracted. Thirdly, according to the characteristics of radar echo, this paper presents a method based on compressed sensing. In the first step, we achieve sparse data by sparse decomposition. Then we choose Gauss random matrix to measure radar echo to obtain the feature vector. The dimension of the feature vector is low, which is consisting of a small amount of echo measurement, but it maintains the structure of the original signal and enough information.Classifier design is another key step to achieve target recognition. Classifiers are trained based on the above various features. In order to take full advantage of the orthogonal and complementary between the various classifiers, this paper designs a parallel combined classifier to play the advantages of each classifier to obtain a high recognition rate.
Keywords/Search Tags:Low Resolution Radar, Target Recognition, Feature Extraction, Combined Classifier
PDF Full Text Request
Related items