Research On Adaptive Resolution Image Reconstruction Algorithm

High-resolution images can provide detailed information,while low-resolution images can greatly reduce the amount of data sampling.The balance between more details and less data sampling has always been an urgent problem in the study of multi-resolution imaging.Aiming at the automatic selection of high-resolution region and the fusion of multi-resolution measurement results,a multi-resolution image reconstruction algorithm based on the region of interest(ROI)is proposed,the validity of the algorithm is verified on the depth image imaging system and the single-pixel imaging system.The adaptive depth image imaging system starts from structural light illumination and focuses on the multi-resolution depth image reconstruction algorithm based on the selection of the projection template in the region of interest.The ROI in the scene of depth imaging system is preliminarily estimated by the region of interest detection algorithm,and only the region of interest was measured many times in a sequence of projected fringe waves.The multi-resolution depth image is obtained by the calibration and triangulation of structured light depth imaging system,which solves the problem of data explosion caused by the direct ratio of measurement precision and measurement times in the traditional depth image acquisition.The single-pixel imaging system obtains low-resolution images through lowresolution template sampling,which determines the region of interest.Gradually reduces the range of the ROI while gradually improving the resolution of the region of interest,reconstructs images through wavelet transform to solve the problem of multiresolution image fusion.The low-resolution ROI was obtained by using the significance detection algorithm,taking it as the initial contour of the level set,then the high-resolution ROI was formed through iteration of the level set algorithm.The images with low,medium and high resolution are reconstructed by inverse wavelet transform.Experimental results show that the wavelet reconstruction algorithm based on the level set and hierarchical region of interest algorithm realizes image reconstruction with three resolutions,and can achieve good multi-resolution reconstruction results under different measurement compression rates.