Research On Terahertz Reflection Imaging And Image Enhancement Algorithms

In recent years,terahertz imaging technology has developed rapidly and been widely used.However,due to current technical limitations,terahertz imaging faces problems such as low image contrast and lack of local detail information.Improving the hardware conditions to improve the imaging quality has a long cycle and high cost,and using image processing technology to solve this problem is simple and practical,and has gradually become the preferred method.Based on the basic characteristics of terahertz images,this paper uses different algorithms to filter and enhance terahertz images.Combined with the current research hotspots,this paper integrates deep learning ideas and trains neural networks to enhance terahertz images.Compared with traditional algorithms,this method has better performance and higher image quality after processing,which provides new ideas for terahertz image enhancement processing.The main research contents of this article are as follows:1.In order to solve the actual needs of the laboratory,this article completed the optical path design of the terahertz reflection imaging system on the basis of understanding the terahertz imaging methods and principles,and successfully built the reflection imaging system using existing instruments and equipment,and used a variety of samples to carry out Imaging experiment.The imaging results show that the imaging results of this system are good,the resolution can reach 200?m,but the local details and contrast of the resulting image need to be improved.2.In order to solve the problem of low terahertz image quality,this paper systematically analyzes the characteristics of the terahertz image and the factors that affect the imaging quality.For the noise appearing in the image,two filtering algorithms are used to filter the image.Considering the low contrast of images,gray scale stretching and frequency domain enhancement are used to enhance the contrast of images.The advantages and disadvantages of different algorithms are illustrated by comparison,and the applicable scenarios of different algorithms are described.3.Combining the research hotspots in the field of computer vision,the idea of deep learning is integrated into terahertz image processing.By simulating the characteristics of terahertz images,an image library of a certain scale is established,GAN neural network is trained to learn the mapping relationship between low-quality images and high-quality images,and the trained network is applied to real terahertz images.The experimental results show that compared with the traditional enhancement algorithm,the proposed algorithm can significantly improve the contrast of terahertz image,and the local details of the image are more perfect and the visual experience is better.Which provides a new idea for the enhancement processing of terahertz images.