Research Of Multi-focus Image Fusion Algorithms Based On PCNN In NSCT Domain

Nowadays, with the development of the Internet, networking, big data and other high technology. The application of sensors in the promotion of these new technologies is also changing with each passing day. With its application in various forms in all walks of life, performance sophistication, it extracts the data information has become increasingly diverse. Therefore, the ensuing question is how effective will be the integration of different types of sensors to collect information into a new data which can be easily observed, explanation or description. Because of the redundancy and complementarity between these data, a single sensor or sensor type of the existing is difficult to handle the situation. To solve this problem, this data fusion are driven to develop into a new technology. Image fusion is a branch of data fusion and it is to use a specific algorithm to deal with multiple images in time or space, to obtain more abundant, accurate and reliable information to meet the specific needs. Therefore, the technology in today's rapid development of the information society has been more and more attention. Based on the domestic and foreign research, some existing fusion algorithms have been studied in this paper, the characteristics of non-subsampled contourlet transform (NSCT) and pulse coupled neural network (PCNN) are combined to study how to improve the quality of multi-focus image fusion.This paper studies the main content as follows:1. This paper studied the principles and structure of contourlet transform and NSCT, based on NSP of NSCT and NSDFB two parts filter group's diversity, with specific NSCT fusion rules, using the experiment results and objective evaluation criteria, the best filter group was selected.2. Aiming at the defects of the traditional PCNN algorithm, such as neuron excitation is the image pixel values, the intensity of the linking ? is a reference to the experience value. This paper selects the image pixels in the spatial frequency as the excitation of PCNN neurons, and chooses the information of the image pixel as the linking intensity ? of PCNN relationship to improve these defects. The fusion effect of the method is verified by the final fusion results and the objective evaluation standard.3. In the design of image fusion rules, decomposition of the original image by NSCT transform, the rule of low frequency and high frequency subband coefficients are different methods respectively. The method proposed in this paper, the sub bands were obtained from the SF as compound PCNN (dual channel PCNN and PCNN) incentive, which is low frequency sub-band input into PCNN and high-frequency sub-band input into the dual channel PCNN. The fusion experimental results of visual contrast and considerable evaluation standard data show that the algorithm proposed in this paper has obvious advantages over the traditional methods in the data of the human eye's visual effect and the objective evaluation.