A New and Effective Multi-Focus Image Fusion Algorithm Based on Wavelet Transforms and Neighborhood Features
Lei Guo,Gong Cheng,Tianyun Zhao
DOI: https://doi.org/10.3969/j.issn.1000-2758.2011.03.024
2011-01-01
Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University
Abstract:Aim. The introduction of the full paper reviews a number of papers in the open literature and then proposes what we believe to be a new and relatively more effective algorithm, which is explained in sections 1 and 2. The core of section 2 consists of: (1) we use wavelet transforms to perform the multi-scale decomposition of source images, thus obtaining their low-frequency and high-frequency sub-images respectively; (2) we apply the neighborhood normalized gradients of pixels to fusing the low-frequency sub-images so as to obtain their low-frequency fusion coefficients and apply the neighborhood variances of pixels to fusing high-frequency sub-images so as to obtain their high-frequency fusion coefficients; (3) we use the inverse wavelet transforms to perform the wavelet reconstruction of the fused sub-images, as shown by the block diagram in Fig. 1, thus obtaining their fused image. Section 3 did two things to compare the fusion effects of different image fusion algorithms: (1) we did experiments on the fusion of two grey images and two color images respectively; the experimental results, given in Figs. 2 and 3, and their analysis show preliminarily that our multi-focus image fusion algorithm can effectively improve the quality of fused image in terms of its definition and contrast; (2) we performed the objective evaluation of the performance of our image fusion algorithm; the evaluation results, given in Table 1, and their analysis also show preliminarily that our image fusion algorithm can reduce the number of decomposition layers and the computation load, thus being superior to other algorithms in terms of root mean squares error, entropy and peak signal to noise ratio.