Font Size: a A A

The Application Of Spatial Binary Nonlinear Correlation In Optical Pattern Recognition

Posted on:2007-04-20Degree:DoctorType:Dissertation
Country:ChinaCandidate:Q WangFull Text:PDF
GTID:1118360185468088Subject:Optics
Abstract/Summary:PDF Full Text Request
With the development and extensive application of Spatial Light Modulator(SLM) and CCD, many attentions have been payed to the optoelectronic hybrid pat-tern recognition systems. The ?exibility and programmability of electronic systemmake various nonlinear technologies available in the field of optical pattern recogni-tion conveniently. Spatial Binary Nonlinear Correlator (SBNC), here we discuss, isa kind of nonlinear optical pattern recognition systems with excellent performance.In SBNC, input and reference images are firstly decomposed into a set of binary greyslices. In general, SBNC appears as linear summation of a set of linear correlations be-tween these binary slices. According to the style of image decomposition, SBNC maybe classified into two kinds: Sliced Orthogonal Nonlinear Generalized Correlation(SONG Correlation) based on sliced orthogonal nonlinear generalized decompositionand Morphological Correlation (MC) based on threshold decomposition. Comparedwith common Linear Correlation (LC), the output performances of SBNC are far bet-ter in either discrimination or noisy robustness. However, as a novel nonlinear corre-lator, SBNC still has many drawbacks. There exist wide sidelobes around correlationpeaks of SONG correlation and MC, which would affect the strict locating of wantedobject seriously. The improvement they make in the noise robustness is still confinedin a relatively small noisy intensity range. Moreover, the correlation signal of MC ismuch weaker than that of LC. Through introducing some other highly developed filtertechniques into SBNC, further improving the performance of SBNC is just the themeof my dissertation.Phase-only Filer (POF) and Fringe-Adjusted Filter (FAF) are two kinds of filter-ing techniques widely used in pattern recognition with good output performance. Weintroduced them into SONG correlation, presenting POF based on SONG Decomposi-tion (PBS) and SONG Decomposition Based Fringe-Adjusted Joint Transform Corre-lation (SBFJTC), and their optoelectronic hybrid implementation setups. In PBS andSBFJTC, the sub-linear correlations in the definition of SONG correlation are sub-stituted by POF and fringe-adjusted joint transform correlation respectively. Simula-tion results show that introduced POF and FAF perfectly smooth away the extraneous...
Keywords/Search Tags:Spatial Binary Nonlinear Correlator, Sliced Orthogonal Nonlinear Generalized Correlation, Morphological Correlation, Phase-Only Filter, Fringe-Adjusted Filter
PDF Full Text Request
Related items