Abstract
Previous Convolutional Neural Networks (CNNs) based multi-focus image fusion methods rely primarily on local information of images. In this paper, we propose a novel deep network architecture for multi-focus image fusion that is based on a non-local image model. The motivation of this paper stems from local and non-local self-similarity widely shown in nature images. We build on this concept and introduce a recurrent neural network (RNN) that performs non-local processing. The RNN captures global and local information by retrieving long distant dependencies, hence augmenting the representation of each pixel with contextual representations. The augmented representation is beneficial to detect accurately focused and defocused pixels. In addition, we design a regression loss to address the influences of texture information. Experimental results demonstrate that the proposed method outperforms the state-of-the-art methods, both qualitatively and quantitatively.
Highlights
Most image systems, such as digital cameras, have a limited depth-of-field and, as a result, only the objects in the depthof-field are sharp, while others are blurred
The proposed fusion method is compared with eleven state-of-the-art multifocus image fusion methods, including the nonsubsampled contourlet transform and sparse representation (NSCT-SR) [8], the multi-scale weight gradient fusion (MWGF) [44], the image matting (IMF) [18], the dense SIFT (DSIFT) [25], the convolutional neural network (CNN) [27], selfsimilarity and depth information (SSSDI) [26], boundary finding (BFMM) [45], cross bilateral filter (CBF) [46], the convolutional sparse representation (CSR) [16], the multiscale convolutional neural network (MADCNN) [35] and IFCNN [37]
Considering the above comparisons based on visual perception and evaluation metrics together, our proposed fusion method outperforms the other eleven state-of-the-art algorithms for multi-focus image fusion
Summary
Most image systems, such as digital cameras, have a limited depth-of-field and, as a result, only the objects in the depthof-field are sharp, while others are blurred. Many researchers have designed various algorithms called multifocus image fusion to integrate several images of the same scene into an all-in-focus image. Many algorithms have been proposed for multi-focus image fusion. These algorithms can be categorized into four classes: transform domainbased methods, sparse representation-based methods, the CNN-based and spatial domain-based methods. Some representative transform domain-based methods include the Laplacian pyramid (LP) [1], the morphological pyramid (MP) [2], the discrete wavelet transform(DWT) [3], the dual-tree complex wavelet transform (DTCWT) [4], the curvelet transform (CVT) [5], the non-subsampled contourlet transform (NSCT) [6], [7] and the sparse representation (NSCT-SR) [8].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
