Font Size: a A A

Research On Multi-targets Serial-number Sustainment On Optical Measurement Device

Posted on:2007-03-02Degree:DoctorType:Dissertation
Country:ChinaCandidate:M J WangFull Text:PDF
GTID:1118360185989746Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
In this paper, multi-targets measurement in shooting range experiment is studied, and keep the targets'number is the essential, the problem that keeping the serial number after targets intersect is settled as a keystone.Firstly, targets measurement have the quality that the SNR is quite low, the traditional edge dected methods can't extract the targets effectively, biorthogonal wavelet possess the characteristic of close support and symmetry, It can attain the best effect of filter. The capability of extracting the faint target is compared among the 3.1ψbiorthogonal wavelet, haar wavelet and sobel, experiments show that 3.1ψbiorthogonal wavelet has the best capability of extracting the faint target, and is not sensitive to noise.Secondly, in the process of keeping the targets'number, in order to overcome the localization trait that probability data association deal with the multi-target, and the miscellaneous trait for joint probability data association, a kind of method of multi-parameters probability data association is put forward in this paper. It introduce the following factors, involving the likelihood of space, the likelihood of lighteness, the likelihood of shape, figure out the integrated posterior probability of each candidate measurement found in a validation gate, hereby, different candidate measurement is put different power coefficient to update the target's state. When the quantity of targets increase, the multi-parameter probability data association's reliability for keeping target's number is close to that of JPDA, and the calculation has great decline compared to that of joint probability data association.At last, the problem of targets intersection each other is discussed in-depth, a kind of track forecast method that can wipe off the noise is put forward. It eliminate...
Keywords/Search Tags:Biorthogonal wavelet for 3.1ψ, Multi-parameters probability data association, Generalized Hough transform, Anti-noise forecast method
PDF Full Text Request
Related items