Research On Underwater Imaging Method Based On Block Compressed Sensing

The ocean is rich in resources and energy.Oceans must be observed and studied before they can be used.In military,geographic,biological,offshore oil,port engineering and other fields,it is very important to take photographs of underwater objects.Traditional underwater optical imaging technology is increasingly difficult to meet the needs.The introduction of Compressed Sensing to inject fresh blood into underwater optical imaging technology will have a revolutionary impact on underwater imaging.Block compression sensing will further improve the underwater image quality.In this paper,the underwater block compression sensing imaging methods are studied in depth,and the main work is as follows:(1)In this paper,the traditional block compression perception algorithm uses the same sampling rate for all(same scale)image blocks,resulting in waste of sampling resources,block effect,and poor quality of reconstructed images.Overlapping blocking and adaptive sampling are introduced into block compression perception to resolve the blocking effect.They are studied in both spatial and wavelet domains.First,in the spatial domain,pre-estimated images are obtained by presampling,and then each image block is given an appropriate sampling rate based on its pre-estimated image characteristics.An adaptive overlapping block compression sensing algorithm with sampling rate is implemented in the spatial domain.Secondly,in the wavelet domain,the low frequency coefficients are used to obtain the preestimated image.Based on the pre-estimated image block characteristics,the adaptive allocated sampling rate is applied between the blocks of the same scale wavelet coefficients,and the sampling rate is implemented in the wavelet domain to accommodate the multi-scale overlapping block compression sensing algorithm.Finally,the adaptive sampling rate based on gray entropy and edge information is studied.In addition,the correlation of gray level co-occurrence matrix of image is introduced as the criterion,and the sampling rate is adaptively sampled.The experimental results show that only a small amount of computational complexity is added compared with traditional block compression perception,and the quality and visual effect of the reconstructed image are significantly improved.Compared with the adaptive sampling method based on gray entropy and edge information,the adaptive sampling method based on gray level co-occurrence matrix presented in this paper has better reconstruction performance.(2)In the underwater environment of rivers and lakes,there are a large number of suspended particles,which have strong absorption and scattering effects.In this paper,an improved underwater imaging algorithm based on GLCM-MS-OBCS is proposed.Unlike the traditional multi-scale block compression sensing algorithm based on wavelet transformation,the low frequency subbands are simply preserved.In this paper,the low frequency subband images are equalized by gray histogram and the high frequency subbands are thresholded.The experimental results show that the performance of the proposed algorithm is better than that of existing algorithms.