Research On The Data Acquisition Technology Of Snapshot Imaging Spectrometer

As a kind of multidimensional information acquisition technology, spectral imaging technology has been widely used in agriculture detection, ocean exploration, environmental monitoring and other fields. Because of obtaining spectral data cube by one exposure and being hand-held, snapshot imaging spectrometer, the new spectral imaging equipment, has a great application prospect. However, the classic spectral image software mainly processes off-line spectral data, and can’t work with the real-time snapshot imaging system. To some extent, this kind of software restricts its development. Therefore, taking our laboratory’s spectral imager prototype as the research platform and the data processing methods as the research object, this paper finally designed the processing software by integrating the real-time acquisition, display and processing of spectral data.Firstly, from the development of spectral imaging technology, the paper discusses snapshot imaging spectrometer’s principle advantages and applications, introduces the development trend of image processing tools, and points out the necessity and importance of the research work in this paper.Secondly, the corresponding solutions to some key problems during processing spectral data are studied. This paper also discusses the transformation model, procedures and algorithms of the spatial position registration in detail. Meanwhile, adopt some algorithms based on pixel gray values, such as normalized cross correlation(NCC), sequential similarity detection algorithms(SSDA), normalized mutual information(NMI), to do the manual registration experiments of spectral images. These results show that: NMI has the highest accuracy but with the longest time; SSDA has the most unstable performance; NCC has the middle precision but with a fast algorithm. In view of this, we finally select NCC for real-time registration.Again, considering the human impact of manual registration process, the automatic registration method based on NMI is proposed. By applying the PSO and Powell for registration optimization, we can draw the following conclusions: 1) The automatic registration with PSO or Powell, has a better performance than the manual NCC and has a faster speed than the manual NMI; 2) The automatic registration with Powell has a higher speed and more stable performance than the registration with PSO; 3) PSO and Powell have the common disadvantage, namely that the accuracy of registration has not yet reached 100%.Then, the data processing software for the snapshot spectrometer system in the Windows platform is designed, using Visual Studio 2010 compiler environment, a single document framework of MFC, and C/C++ language. According to snapshot imaging spectrometer’s principle, this paper implements the spectral data’s real-time acquisition, gain and exposure-time adjustment, color correction, position registration, data view, false color composition, spectral curves’ extraction and comparison, image enhancement and so on. At present, the software has run successfully for about half a year in the laboratory.Finally, the paper does a comprehensive summary for the work in this paper, including some existing deficiencies, and makes some suggestions for the next work. Moreover, discuss the future development direction of the snapshot spectral data processing software.