Color Adaptive Compression Ghost Imaging

Colored imaging plays an important role in the fields of detection and recognition.Currently,however,there is little research aimed at reconstructing color images.The conventional colored imaging method is to recover the red(R),green(G),and blue(B)components separately using the methods for gray-scale images,such as compressed sensing(CS).However,on the one hand,CS has a huge computational overhead,and the time consumed increases exponentially with the size of the image to be recovered;on the other hand,if one reconstructs the red,green,and blue components separately,the visual effect of the reconstruction image is far from satisfactory since the strong correlation among the three components is ignored.Considering the existing methods have some shortcomings in that the reconstruction time is too long,and the visual effect of the reconstruction image is far from satisfactory.We proposed three colored adaptive compressed imaging methods step by step in this paper:1.EWT-ACS RGB.Adaptive compressed sampling based on extended wavelet trees(EWT-ACS)is introduced in colored imaging,which is exploited to reconstruct the R,G and B components separately.This method reconstructs the images of any size quickly,without any computational overhead,and thus improves the imaging speed significantly.2.EWT-ACS RGBY.In order to improve the image quality further,this method samples the luminance imformation of the object to be recovered,which is represented by Y component.EWT-ACS RGBY improves the quality of the color images effectively and makes the details of the images more clear.3.EWT-ACS YUV.Benefiting from imaging in YUV color space,the proposed method adequately exploits the sparsity of the U,V components in the wavelet domain,the interdependence between luminance and chrominance,and human visual characteristics,and avoids modeling the correlation among the R,G and B components.EWT-ACS YUV greatly reduces the measurements required and improves the image quality a lot.In order to check the validity of the proposed methods and the numerical simulation results,an experiment was performed.The results show that the proposed methods greatly reduce the measurements required and offer better image quality,especially EWT-ACS YUV method.