| Optical imaging is one of the main ways for human beings to obtain information.In recent years,with the development of optical imaging techniques,a new type of computational imaging technique called single-pixel imaging has received widespread attention.At present,this technique has shown broad application prospects in many fields.C ompared with the traditional optical imaging system based on pixelated detector arrays,this imaging technique based on a single point detector has significant advantages such as wider photosensitive spectrum,more sensitive to light and higher frequency response bandwidth.However,in order to obtain the image of the object,single-pixel imaging usually needs to use the single-pixel detector to measure a large number of light intensity signals in sequence,which is time-consuming;Moreover,if the relative motion between the object and the imaging system occurs,the final restored image will deteriorate.In view of these problems,this paper mainly carries out two aspects of research work.On the one hand,on the basis of previous studies,this paper proposes a sparse Radon single-pixel imaging technique to solve the problem that the restored image quality is poor under the condition of a low sampling rate.On the other hand,two new single-pixel imaging methods are proposed to greatly alleviate motion blur when single-pixel imaging is used to image the fast-moving object.Specifically,the main research work of this paper can be summarized as follows:(1)After Radon single-pixel imaging obtains a series of detection signals by the single-pixel detector,the Radon spectrum of the object needs to be calculated first,and then the inverse Radon transform is performed on the Radon spectrum to restore the object image.Because there doesn’t exist an exact implementation of the inverse Radon transform in the discrete case,the filtered back projection algorithm is usually used to realize the discrete inverse Radon transform.However,when the sampling rate is low,the image restored by this method will appear serious fringe artifact noise.In order to solve this problem,this paper proposes a sparse Radon single-pixel imaging technique,which can effectively reduce the number of samples and ensure a relatively high quality of restored images.The proposed method uses the designed two-dimensional projection patterns to perform sparse angle sampling,and then uses the compressive sensing algorithm to invert the obtained data to restore high-quality images.From the results of computer simulation and experimental results,it can be found that the quality of the restored image of the proposed method is significantly better than that of the filtered back-projection algorithm at the same sampling rate,and it can still restore a high-quality image at a low sampling rate.In addition,considering the wide application of Radon transform in medical imaging,non-destructive testing,object classification,recognition,and so on,the proposed method also has application prospects in these fields.(2)Generally,in order to capture high-quality object images with single-pixel imaging,it is necessary to use a large number of illumination patterns to spatially modulate the light and project the modulated light onto the target object;and the single-pixel detector will sequentially measure a large number of light intensity signals corresponding to the illumination patterns,then these measured signals are used to restore the image of the target object.When imaging a static object,this imaging mechanism can effectively restore high-quality object images.When imaging a moving object,if the moving object can be regarded as a relatively static state within the time required to obtain sufficient light signals,then a clear image of the object can also be taken.However,with the increase of the object’s motion speed,if the object’s motion speed exceeds the corresponding boundary conditions within the time required for single-pixel imaging to obtain sufficient detection signals,the traditional single-pixel imaging strategy for static objects will lead to more and more serious motion blur in the reconstructed image.To cope with this problem,this paper presents an anti-motion blur Radon single-pixel imaging technique.The proposed method uses the single-pixel imaging system to obtain the one-dimensional projection curves of the moving object,and then uses these projection curves to calculate the object position information and correct the Radon spectrum according to these position information,and finally uses the corrected Radon spectrum to restore the image of the moving object.Computer simulation and experimental results show that the proposed method can greatly alleviate motion blur caused by motion and effectively restore the image of moving objects without prior information such as velocity and trajectory.(3)Although the motion blur caused by the motion of the object can be effectively alleviated by restoring the image of the moving object with the corrected Radon spectrum,when the sampling rate is low,the restored image will inevitably appear artifact noise because the filtered back projection algorithm is used for image restoration,which will lead to the reduction of the quality of the restored image.Although the image restored by compressive sensing is usually of high quality when imaging static objects,for moving objects,the compressive sensing algorithm cannot correct the impact caused by object motion,and it is difficult to restore high-quality images of objects directly using the compressive sensing algorithm.In order to solve these problems,this paper presents a compressive Radon single-pixel imaging of the fast-moving object technique.This technique removes the motion interference in the measurement signal by calculating the position information of the object motion to obtain the calibration signal,and then takes the calibration signal as the observation value and uses the compressive sensing algorithm to restore the image of the moving object.From computer simulation and experimental results,it can be found that the proposed method can not only greatly alleviate motion blur but also reduce the background noise of the restored image,which effectively improves the quality of the reconstructed image.(4)In addition,the application of Radon single-pixel imaging in edge detection is also studied.By translating the two-dimensional projection pattern,and then using the measured signals before and after translation to calculate the gradient Radon spectrum.Finally,based on the gradient Radon spectrum,the edge image reconstruction,linear edge positioning,and edge image correction are studied respectively. |
