Objective: Linear feature extraction is very important in image processing. In ICT cubic data

line is one of the important features. The lines’ extraction efficiency will directly affect the performance of pattern recognition and classification. Method: Finite line integral transform (FLIT) is a new method based on the concepts of multiscale geometric analysis (MGA). Using a series of fixed templates in different scales

FLIT gets a series of decomposition images containing the original image’s information at different scales and directions. FLIT is invertible and non-wrap

which is very successful in image denoising

line extraction and recognition when applied to images with line singularities. Based on the concept of 2D-FLIT

this paper generalize it to the case of 3D

attaining the formulas of 3D-FLIT and inverse 3D-FLIT. In this paper

in order to extract linear feature of ICT cubic data

the authors present two methods based on FLIT (2D-FLIT or 3D-FLIT)

including other image processing

such as image fusion and morphological. The first strategy is the idea of 3D-FLIT applying on the ICT cubic data directly for detecting linear features. The second strategy is the idea of 2D-FLIT applying on each slice in some direction of the ICT cubic data for detecting linear features. Then different directions’ images are fused for obtaining the linear features. Finally morphology image processing is used for extracting useful components of the linear features. Result: Numerical experiment results prove the efficiency of the proposed methods compared with wavelet method. Conclusion: Comparatively speaking

because the first strategy avoids slice dividing and saves the voxel relation between adjacent slices

its extracting linear features’ result is better and full than the second strategy. But

the second strategy is simpler and has less computation time. So

each of the two strategies has its own strong point.