The Research Of Multi-energy X-ray Image Stack Fusion Algorithm

The X-ray imaging of complex structural components isn’t normal since the imagingdevice’s dynamic range is not enough. Multi-energy X-ray imaging technology is presentedto solve the problem and it can extend imaging system’s dynamic range. The image containedcomplete information of complex components’ structure is gained by multi-energy X-rayimage stack fusion. The fusion method is weight fusion and the weights are computed bygray-voltage-thickness model which is created with ideal monoenergetic X-ray. The weightsneeds to dynamical adjustment with manual intervention since the polyenergetic X-ray inpractice, which limited the application of the technology. This paper proposed fusion methodsto improve the self-adaption of fusion and physical matches between fusion image and thecomponent.Firstly, the X-ray images fusion is based on principle component analysis (PCA). Thefusion image is obtained through the principle components of X-ray image stack. So theweights are self-adaption. Secondly, the X-ray image stack fusion is based on grayconsistency. A fusion model is created with the distance of images. The weights can be got bysolving the model. This makes the weights have physical meanings. Thirdly, the X-ray imagestack fusion is based on useful areas. The useful area of every X-ray image is automaticallyselected by an evaluation model for image quality. This method avoids influence of lowerquality local image to fusion image and can be automatically stopped by recursive fusionwhen the information of fusion image is complete. The experiment shows that the threefusion method all can obtain fusion image with complete structure information of complexcomponents, but the fusion image with last method is much closer to actual image thanprevious methods.