Research On Snapshot Multispectral Imaging Technology Based On Smartphone

Non-invasive multispectral imaging of physiological changes in biological systems has been widely used in various research fields,including medical and biological applications.However,traditional methods require key optomechanical components(i.e.,mechanical filter wheels,gratings,prisms,or liquid crystal tunable filters)to capture full-spectral information.Multispectral imaging systems employing mechanical filter wheels are bulky and fast.Slow;multispectral imaging systems using gratings or prisms cannot directly complete imaging;multispectral imaging systems using liquid crystal tunable filters have a longer response time.And this component makes the whole system expensive.Therefore,there is a need for a convenient and relatively low-cost snapshot system.In the past ten years,smartphones have been widely used in people’s daily life.Therefore,in this paper,the smartphone camera is used to capture images,and spectral reconstruction algorithms are used to convert the acquired images into multispectral images.Firstly,this paper studies five commonly used multispectral reconstruction algorithms,namely Wiener estimation method,optimal Wiener estimation method,adaptive Wiener estimation method,polynomial regression method and root polynomial regression method.And under the actual environment and lighting conditions,the above-mentioned multi-spectral reconstruction algorithms are experimentally studied,and it is found that the reconstruction accuracy of the five multi-spectral reconstruction algorithms is insufficient.Therefore,an innovative and improved spectral reconstruction algorithm is proposed,namely the Wiener estimation reconstruction method based on the combination of polynomial regression and root polynomial regression.Then,this paper selects Color Checker Classic 24 colors and Color Checker DIGITAL SG 140 color standard color cards and self-made fluorescent color cards as color standards.Obtain the initial spectral data of the color swatch with a spectral analyzer.In the same environment,the color card image was taken with a smartphone,and an empirical study on the above six spectral reconstruction methods was carried out.The advantages and disadvantages of the six reconstruction methods are compared and studied in terms of root mean square error and relative error.The experimental results show that the Wiener estimation reconstruction algorithm based on the combination of polynomial regression and root polynomial regression proposed in this paper is an effective spectral reconstruction algorithm.The average spectral root mean square error and average relative error of the reflection spectrum reconstruction of the proposed method are reduced to 0.0233 and 0.0689,respectively,which are higher than the above-mentioned commonly used spectral reconstruction algorithms in terms of spectral reconstruction accuracy.Finally,the paper introduces the application of the proposed multispectral reconstruction algorithm in the field of biomedicine.In this paper,the Wiener estimation reconstruction method based on the combination of polynomial regression and root polynomial regression is used to reconstruct the acquired color image into a multispectral image,and then the weighted subtraction between the bands is used to extract the specific chromophores(mainly including hemoglobin and The absorption information caused by melanin)and the autofluorescence spectra of porphyrins produced by bacteria to achieve bacterial localization,realizing skin tissue chromophore extraction and separation and high contrast and high signal-to-noise ratio identification of skin and oral bacteria.