Super-Resolution Imaging In Atomic Force Microscope

Atomic force microscope(AFM)is a powerful nano scale instrument,which can get the true surface morphology of samples.It is widely used in materials science,chemistry and biomedicine.AFM can not only work in liquid or vacuum environment,but also scan nano scale conductors,semiconductors and insulating materials,which makes AFM more applicable.Although the vertical resolution of AFM can reach0.1nm,the lateral resolution of AFM is not high due to the influence of probe and sample morphology.In addition,AFM imaging is a complex and time-consuming process due to the point by point scanning mode and positioning accuracy.Therefore,most of the images are of poor quality,and the original AFM images are low-resolution.However,it is very important to obtain high-resolution images for nanoscale measurement and imaging.Although the image resolution can be improved by improving the hardware facilities,their enhancement is very limited and the cost is very high,which can not meet the needs of scientific research.Therefore,super-resolution technology is one of the most effective way to solve the problem of AFM image,which can use software algorithms to obtain high-quality AFM image without limited hardware facilities.According to the shortcomings of different super-resolution technologies,corresponding improvements are made to achieve higher quality AFM super-resolution imaging.In view of the fact that AFM image has a lot of texture details and the traditional interpolation algorithm does not consider the image contrast information,a contrast-oriented image super-resolution method is used to process the edge details.Furthermore,the edge-oriented interpolation algorithm is improved by reducing the computational complexity in the original iteration process and introducing accuracy higher iterative back projection algorithm further improves the quality of image edge.In order to solve the problems of slow iterative speed and poor image quality of traditional reconstruction algorithm,compressed sensing theory is applied to AFM image super-resolution imaging,and a special measurement matrix and TVAL3 reconstruction algorithm are adopted to shorten the scanning time of AFM and significantly improve the image quality after super-resolution.Convolution neural network is used to realize AFM super-resolution imaging.In order to solve the problems of long time and high cost of large AFM image database,the algorithm based on convolution neural network is improved,and the data set is enhanced by adaptive histogram equalization.The deep learning model is further improved,and the quality of image reconstruction is improved accordingly.The edge of image interpolation,the details based on reconstruction and the data set of deep learning are improved respectively,and the subjective and objective image quality evaluation indexes are used to evaluate the proposed super-resolution algorithms.The experimental results show that the quality of AFM image is effectively improved,which proves the feasibility of the research method.