Research On Interpolation Method And Application Of Sequence Images

In the fast-developing digital age,digital images have become the main medium for transmitting visual information,and people's requirements for image quality are gradually increasing.Therefore,in recent years,digital image processing technology,especially image enhancement technology,has been rapidly developed and widely used in many fields.For example,in order to make the video more smooth,it is necessary to synthesize the middle video frame to increase the frame rate;in order to analyze the atmospheric and ground conditions,it is necessary to magnify and deblur the remote sensing satellite images;in order to detect cases,it is necessary to super-resolution processing of surveillance images;in order to locate the lesion more accurately,it is necessary to segment and super-resolution reconstruction operation of medical images,etc.These image processing processes to improve image clarity and resolution all need to use image interpolation techniques.Image interpolation is the process of estimating the pixel values of unknown sampling points using known sampling points of an image,aiming to turn a low-resolution image into a corresponding high-resolution version to improve the visual effect of the image.Depending on the direction of resolution enhancement,the interpolation process can be carried out in the spatial and temporal domains of the image sequence.Classical interpolation algorithms focus on spatial resolution enhancement,turning a single low-resolution image into a high-resolution version.Resolution enhancement in the time domain is called inter-frame interpolation,which aims to transform a low-frame-number image sequence into a high-frame-number version to make the image sequence change more naturally and smoothly.It is widely used in the fields of frame rate conversion and biomedical imaging.In the field of video processing,in order to make full use of the limited bandwidth,low-resolution video streams are usually transmitted,so interpolation algorithms need to be used at the receiving end to convert to the corresponding high-resolution video streams to bring a smoother video playback experience.In medical image reconstruction tasks,the inter-layer resolution of imaging data is significantly lower than the intra-layer resolution due to the inherent physical limitations of imaging devices,which poses a challenge to post-analysis processing and clinical diagnosis.To this end,this paper investigates the inter-frame interpolation technique for sequence images,and the following two improved algorithms are proposed for the application fields of video inter-frame interpolation and medical image inter-layer resolution enhancement:The inter-frame interpolation method based on optical flow estimation uses the registration information of two adjacent frames to generate a new intermediate frame,and repair the obtained intermediate frame.First,with the help of CLG-TV optical flow estimation model,the pixel motion between two consecutive images was estimated,for obtaining the pixel-level correspondence between images.Secondly,the velocity vector is scaled with reference to the interpolation position of the target image,and the interpolated image is generated by calculating the velocity field.Finally,the non-local self-similarity between sequence frames is used to repair the pixel missing caused by optical flow estimation.Quantitative and qualitative experimental results on video and CT image datasets show that the method generates high quality intermediate frame images and improves the video frame rate and the inter-layer resolution of medical images to some extent compared with other classical methods.In view of the requirements of medical image processing system for high reliability and real-time,in order to improve the efficiency of inter-layer interpolation operation,this paper proposes an image inter-layer interpolation method based on space geometry polynomial fitting.First,the sampling formula is used to inversely sample the pixels on the known image sequence,and the three-dimensional quadratic polynomial space geometry is locally constructed in the 3×3×3 spatial domain centered on the sampled point.Then,the constructed polynomial space geometry is weighted and averaged to obtain the unit space geometry,and all the unit space geometry are combined.Finally,resample it to generate new pixels to interpolate the intermediate image.The method avoids the complicated process of optical flow calculation,improves the operation efficiency of the algorithm,and the generated interpolation results obtain satisfactory quantitative and qualitative indices.In addition,the method enables interpolation between two consecutive images to generate an arbitrary number of intermediate images,thus effectively improving the axial resolution of medical image sequences.