The Research On Compressive Imaging Method Based On The Guidance Of Prior Knowledge

With the rapid development of information society,the traditional Nyquist sampling law to guide the information processing is facing two major challenges:high-speed analog-to-digital converter A/D and massive data processing and storage.As a new signal acquisition theory,compressive sensing(CS)theory uses sparse or compressible signal.The sampling frequency is lower than Nyquist law get all the information of the signal through non-adaptive linear combination of signal global.Compressed sensing uses only a small amount of acquisition data and reconstruction algorithms to approximately rebuild the original signal.Compressive imaging(CI)as a CS theory in the field of imaging applications,it uses only a single point detector and a lower sampling frequency,which can significantly save the number of photosensitive components,lower the requirements in hardware performance and system complexity compared with the existing imaging technology.Based on the sparseness of natural images,Cl technology can accurately restore images with a small amount of data collected by multiple acquisitions of the measured images.In essence,CI is a computational imaging,the higher resolution of the recovered image is,the more times of measurements and longer the image is restored.The important information of the image is mainly concentrated in the limited target area,and its proportion in the image often does not exceed 30%of the image as a whole,while the background part contains less information without high-resolution imaging requirements.However,the existing CI method uses the same measurement accuracy without distinction on the background and target objects in the image to be measured,and performs high-resolution measurements on the background part while performing high-resolution measurements on the target object.Therefore,this method wasted a lot of computing resources in the background area,which reduced the efficiency of imaging.At the same time,the image information to be measured is susceptible to changes in the intensity of the outside world during CI,resulting in poor quality of the restored image and even the image information being masked by noise.We proposed a CI method based on prior knowledge guidance according the above problems of CI method.Firstly,this method uses the low resolution CI of the image to be measured,and obtains the distribution of the target object and the background in the low resolution image through the visual significance detection method as the prior knowledge.Then,achieving iterative CI only on the target object up to the highest accuracy of the system.The results of each iterative imaging use the visual significance detection method to obtain more accurate target object position and size,while reduces the target object area included in the background information.Finally,the high-precision target image is fused to the background image of the low-precision measurement result,and the final image to be measured is obtained.When the external light intensity changes by the interference,the use of the total light intensity of the measured image measurement results,the CI process to eliminate the deviation between measurement data and the ideal measurement results to a certain extent.The simulation results show that the number of measurements is 209%-30%of the number of measurements of the existing compression-sensing imaging method when the target object size is 10%-15%of the whole image.The accuracy of image restoration can be improved by the method of tracking the total light intensity of the measured image.Based on the above research,the research work of this paper can improve the imaging efficiency and accuracy on the existing CI method.