Multi-Component Decomposition Of 3D Block-Matched Dual-Energy CT With Interlayer Constraint

Dual-energy CT can realize the characterization of base material components under nondestructive conditions according to the different information of the attenuation coefficient of the object under different energy X-rays,but there is an error between the material attenuation coefficient value obtained under the actual conventional CT system and the theoretical value of the attenuation coefficient,which affects the accuracy of the component characterization results.Although the existing research uses the block matching method to add the obtained structural similarity information to the material decomposition,which effectively reduces the pollution of image noise in the material decomposition and improves the material segmentation accuracy,the segmentation accuracy of industrial materials with high density in threedimensional space is still limited.To solve this problem,based on the connectivity and consistency characteristics of interlayer materials,this paper combines block matching and interlayer constraint thought to form a three-dimensional block matching method with interlayer constraint based on interlayer constraint thought,obtains the three-dimensional spatial structure similarity information of pixels,introduces more comprehensive structure similarity information,effectively reduces the impact of noise in the reconstructed image on the decomposition of base materials,and improves the accuracy of material decomposition.Aiming at the problem of inaccurate component representation caused by the error of the attenuation coefficient of high-density material obtained by the actual CT,this paper considers the interlayer material connectivity and introduces the interlayer constraint idea,uses the three-dimensional block matching method with interlayer constraint to add the threedimensional structure similarity information to the pixel’s base material selection,and introduces the multi-material decomposition model to obtain the initial base image,Then the3 D block matching method is used to group the base image with 3D feature constraints,and then the multi-material decomposition is performed again.The experiment shows that the algorithm effectively improves the material recognition ability with a close attenuation coefficient.However,for materials with similar or even equal attenuation coefficients,the differentiation is still limited.To this end,the regionalized 3D block matching is used to obtain more comprehensive structural information,improve the accuracy of the pixel material composition matrix,introduce the multi-material decomposition model with TV regular term to obtain the initial base image,and then use the regionalized block matching to group the base image with 3D feature constraints,and then use the multi-material decomposition with TV minimization to obtain more accurate component characterization diagram,and the experiment proves that The algorithm further improves the integrity and accuracy of component characterization.