In order to obtain a panoramic image which is clearer, and has more layers and texture features, we propose an innovative multi-focus image fusion algorithm by combining with non-subsampled shearlet transform (NSST) and residual network (ResNet). First, NSST decomposes a pair of input images to produce subband coefficients of different frequencies for subsequent feature processing. Then, ResNet is applied to fuse the low frequency subband coefficients, and improved gradient sum of Laplace energy (IGSML) perform high frequency feature information processing. Finally, the inverse NSST is performed on the fused coefficients of different frequencies to obtain the final fused image. In our method, we fully consider the low frequency global features and high frequency detail information in image by using NSST. For low-frequency coefficients fusion, we can also obtain the spatial information features of low-frequency coefficient images by using ResNet, which has a deep network structure. IGSML can use different directional gradients to process high-frequency subband coefficients of different levels and directions, which is more conducive to the fusion of the coefficients. The experiment results show that the proposed method has been improved in the structural features and edge texture in the fusion images.

Introduction

In the field of digital image processing, different imaging devices acquire different information from the same scene. As in the optical lens, the acquired image is not an allfocus image since the limited of the lens depth range. The optical image is only clear in the part of the scene that is focused in the lens range, and the rest is a blurred defocused image. Typically, image fusion is often used to produce good results which is superior to the original image quality [1], [2]. The fused image contains more scene information, which is more suitable for imaging features of the human eye and is also convenient for later computer processing. Therefore, the process of multi-focus image fusion can be considered as a tool for producing high quality result images [3], [4]. In the development of multi-focus image fusion, there are two types of fusion methods, namely spatial domain fusion and transform domain fusion [5]. However, the most important aspect is the design of fusion rules in the image fusion processing. The image fusion methods based on transform domain are a popular and widely in this fields. In transform domain-based fusion algorithm, the multi-scale decomposition of the original images is mainly applied by multiscale transform (MST), and image fusion is performed by using different fusion rules for image coefficients at different scales. In image fusion based on transform domain, the performance of the algorithm is mainly dependent on the choice of transform domain and the design of fusion rules.