Research On Terahertz Image Recognition And Processing Technology

Terahertz imaging technology is a technology that uses terahertz radiation as a signal source to perform imaging.It generally performs imaging by transmitting or reflecting sample information.The core content of this paper is to use digital image processing technology to process terahertz image,and finally achieve the segmentation and recognition of the target area.The main work in this paper includes the following four parts:First introduced the relevant content of the image preprocessing algorithm,contains algorithms for image denoising,threshold segmentation,and edge detection that are most commonly used in image processing.Image smoothing is mainly to eliminate noise,image smoothing is actually low-pass filtering,smoothing can blur the edges of the image.In this paper select the window size of 3 * 3 median filter for smoothing,the processed image not only saves the better edge,but also eliminate most of the noise.Threshold segmentation takes advantage of differences in the gray-scale characteristics of the target area and the background in the image,select the correct threshold by observing the gray-scale histogram of the image or curve fitting image gray-scale histogram,and then separate the target area from its background.Edge detection and recognition through the results of simulation experiments,compared the advantages and disadvantages of different operators,and then an appropriate edge detection operator is selected to improve the accuracy of edge detection.Secondly,realized terahertz image reconstruction,and resolution of the terahertz image is solved,through comparing several high-resolution reconstruction algorithm,finally,the deficiency of the traditional algorithm is compensated by the improved SRCNN algorithm.,and its used in the processing of terahertz images.Then implements terahertz image segmentation algorithm,proposed a segmentation algorithm based on curve fitting,segmented passive terahertz image after image restoration.After de-noising preprocessing of the original terahertz image,by analyzing the gray histogram and fitting the curve,selected seed nodes and growth threshold,in order to achieve image segmentation.Finally realized terahertz image processing and recognition.Combined with the previously described image processing methods and algorithms and processing flow,simulation of terahertz images for cutting tools,fractured traverse and firearms by using Matlab and VC++.Matlab and VC++ simulation results can be seen,this chapter has better image processing results for the terahertz images of cutters,broken wires,and firearms,and can identify terahertz images of knives,broken wires,and firearms.