Spectral Unmixing And Image Mosaic Of Bio-optical Image And Building Of Vein Imaging System

Optical imaging is widely used due to its advantages of non-destructive detection, high sensitivity, high resolution, etc. Due to the biocompatibility of optical band, optical imaging has broad application prospects in the study of disease diagnosis, pathogenesis and development of new medical treatments and so on. Optical image processing is an important part during the application of optical imaging, which has a very important significance in quantifying signal characteristics of research object.The thesis starts from optical imaging and introduces the applications of spectral unmixing and image mosaic technology in optical image processing. In the thesis, we build the experimental system of vein imaging system and introduce the segmentation algorithm for extracting vessel in the infrared vein image.The main contents and innovations can be summarized as the following three points:1) The basic theory of spectral unmixing is summarized, including linear spectral unmixing model, endmember extraction algorithms and the estimation methods of endmember number. The estimation method of endmember number HySime and spectral unmixing algorithm MVSA are combined directly to process the multispectral images, getting the endmember number and isolating the endmember information without any prior knowledge. Using this method, we conduct spectral unmixing experiments on the in vivo small animal multispectral fluorescence images and unmix autofluorescence and signal fluorescence accurately.2) Using image registration and image fusion technology to mosaic pathological microscopic images is proposed. We use a feature points based registration method to mosaic images accurately in spatial position and use image fusion technology to eliminate the seam line, which improves the readability of mosaic pathological microscopic images.3) A combination of median filter and NiBlack algorithm is used to improve the SNR of the image and achieve the dynamic thresholding on this basis of analyzing the characteristics of the infrared vein image, which separates the vessels from hand back information successfully. On this basis, the experimental system of vein imaging system is built, which combines the optical detection, algorithm unmixing as well as in-situ projection of blood vessels and achieve the visual enhancement of the hand back blood vessels.