Algorithm And Experimental Study Of Handwriting Identification Based On Hyperspectral Imaging

Hyperspectral imaging technology has been widely applied in agricultural production,scientific research and even military field.Foreign researchers began to apply hyperspectral imaging technology to material evidence identification in the mid 90s of last century.Hyperspectral imaging technology is to record the reflected light distribution or luminance distribution of a plurality of narrow band monochromatic light.Hyperspectral image is a set of spectral images composed of many monochromatic light images.Each monochromatic image of the hyperspectral image records the information of the luminance distribution of the object in this band.The combination of monochromatic images records the information of the brightness distribution of the object in all monochromatic bands.Therefore,the hyperspectral image has high spectral resolution and continuous band information,and the band is from ultraviolet to infrared.The band range is wide.Hyperspectral image has both spectral and image information.Because different substances have great differences in light absorption and reflection,so the spectral reflectance curves of different brand handwriting will also be very different.Analyzing the original hyperspectral data,we can identify the physical structure and chemical composition difference of the experimental samples.Therefore,the original hyperspectral image can reflect the comprehensive quality of the experimental samples.Hyperspectral imaging technology can also accurately reflect different spectral information between different substances.This paper mainly uses hyperspectral imaging technology in the two aspects of handwriting identification algorithm and experiment.The main contents of this article are as follows:1、The development process of hyperspectral imaging technology is expounded,and some traditional methods of handwriting identification and its application in the field of optics are introduced.2、the structure and hyperspectral imaging principle of hyperspectral sorting instrument,as well as the adjustment of experimental equipment and the collection of experimental data are introduced.The pretreatment contents of experimental data calibration and noise reduction are also introduced.3、we do research on the test and analysis of the erasable handwriting,introduce the preparation methods of the experimental samples,and introduce the processing and analysis of the experimental data.After pre-processing,the experimental data are processed by principal component analysis algorithm,and the final experimental results are obtained through debugging related parameters.The algorithm can reproduce the erased handwriting successfully,and the recognition effect is good.4、The tampered handwriting and masking of 18 kinds of black neutral pens and the tampered handwriting and masking of the blue neutral pen of the 15 brand were detected and analyzed.After pretreatment,we use MNF^[28]transform to reduce data dimension and reduce partial noise.ISODATA^[39]algorithm is applied to classify hyperspectral images.After reverse eye mask processing,background information of experimental data is removed.Combined with PCA^[22]analysis and pseudo color synthesis,a recognition method of forgery and cover up is proposed.