Font Size: a A A

Research On Methods And Applications Of Image Fusion Of FTIR Multi-spectral Microscopic Images

Posted on:2013-11-11Degree:MasterType:Thesis
Country:ChinaCandidate:Y L XingFull Text:PDF
GTID:2248330377459172Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
With the rapid development of the infrared spectroscopy, Fourier Transform Infrared(FTIR) spectroscopy has been demonstrated as a powerful and potential tool formicro-analytical techniques. Compared with the conventional visible-light imagingtechniques, the most prominent advantage of FTIR micro-spectroscopy imaging system is thelow interference and non-destructive in the testing process. Moreover, it provides theopportunity to visualize the distribution of chemical ingredients. Therefore, in recent years,FTIR multi-spectral imaging technology has been applied in many areas, such as food safety,drug testing and cultural identification. However, due to the limitation of the hardware andsoftware conditions, the accuracy of analysis and comparison has been greatly affected.Exploring the FTIR data acquisition and data processing methods has become an importantresearch topic in FTIR application. This study is based on the world’s leading Spotlight-400infrared imaging system to explore multi-spectral image features optimized fusion method,which has a high academic value and practical value in pathological analysis, food, drugtesting and other areas. In this paper, the major research is as follows:Firstly, the basic principle of the FTIR micro-spectroscopy imaging and the samplepreparation methods is briefly introduced. And then, the specification, powerful advantagesand the infrared imaging characteristic of the world’s leading Spotlight-400infraredimaging instrument are described.Subsequently, study on the feature extraction methods about multi-spectral microscopicimages. The fundamental algorithm of PCA is deeply discussed, and by combining thechemistry and the pattern recognition basic method, the new extraction method based onLambert-Bill model is proposed. The experimental results demonstrate the effectiveness ofthis method.Then, multi-spectral microscopic fusion algorithm is discussed. In this thesis,Bi-dimensional Empirical Mode Decomposition (BEMD), which is the current populardecomposition method, is used in the multi-spectral image fusion. According to the need ofmulti-spectral microscopic image fusion, taking images’ decomposition speed andcharacters into account, some important issue is discussed in details, including extracting theextrema points, creating envelope, sifting stop criteria and the border points. At last, takingfull account of the detail information and the spectral features, the decomposed images arefused. Experiment shows that, compared with the wavelet transform method, the new method is not only self-adaptive, but also strengthens the spectral information and detail information.Finally, in order to fuse the FTIR multi-spectral images collected by the instrument, wepropose a novel and efficacious technique using PCA and BEMD. From the new perspective,a combination of the infrared spectral analysis and the multi-spectral digital image processing,combined with the multi-spectral optimization and fusion, the interest region to focus on theinterest areas on the FTIR multi-spectral microscopic images is located and the fusedinformation is got, which has rich information and highlight interest areas. Moreover, as isshown in experimental results related with rabbit artery sample and wheat kernel sample, thenew algorithm is effective.
Keywords/Search Tags:FTIR micro-spectroscopic imaging, feature extraction, principal componentanalysis (PCA), multi-spectral image fusion, bi-dimensional empirical mode decomposition(BEMD)
PDF Full Text Request
Related items