An efficient biomedical cell image fusion method based on the multilevel low rank representation

Medical image fusion is an effective machine learning technique for enhancing the quality of biological data. The fusion approach works by integrating the features of different medical observations into a joint representation matrix. This increases the discriminative power of the coefficient matrix. Which improves the appropriate identification of morphological sub-structures. In the past, several such techniques have been proposed but most of them have performed the single level fusion of the medical data. Which, does not extract essential features appropriately. Also, these techniques degrade the pixel quality of the images. We have addressed the given problems by adopting an efficient multilevel cell image fusion method. It performs the linear image decomposition and fusion of the principal and salient features. In which, the salient patterns are fused by the nuclear norm approach and the principal components are fused with the weighted average technique. The experiments were performed on the different types of cell imaging datasets. In terms of qualitative and quantitative comparison, the adopted approach has outperformed most of the recent methods.