Study On Image Fusion Based On NSCT And Regional Characteristics

As the development of science and technology, especially in computer technology, the data volume is much bigger and data process is much more complex.Multisensor system is a typical example.In comparison with the single sensor system,multisensor system can receive mass data from various different sensors,some of data are repetitive or futile, a lot of data are not suitable for understanding and subsequent processing.Multisensor system need to process a lot of data comprehensively to extract useful data eliminate redundancy data as far as possible.Data fusion technology can solve this problem efficiently,image fusion is an important part of data fusion,image fusion technique can form a clearer,more complete and accurate image by means of integration in accordance with certain criteria by processing the image data from different sensors.With the rapid development of image fusion technique,many efficient image fusion algorithm are proposed.In this paper, an image fusion algorithm based on NSCT transform is proposed.This paper introduces in detail the image fusion technique and makes relating analysis and comparison,research status and main accomplishments of image fusion technique are summarized in detail.The key technologies and basic theory of image fusion were described in detail.Image fusion technique from different angles was compared and analyzed after the analysis and comparison.In this paper,we take NSCT as multi-scale geometric analysis tool of image fusion. The advantage and principle of NSCT were compared with other multi-scale transformation.This paper analyze shortages of traditional fusion rules and design a fusion rule based on NSCT and pixel region.low-frequency and high-frequency coefficients are fused with different fusion rules. For the low-frequency coefficients,we use an fusion rule based on region variance and region energy,this fusion rule overcome the shortcoming of pixel fusion rules in local properties of the image and eliminate traces of image fusion.The high-frequency coefficients are fused by the combination of region contrast and region energy, this fusion method can not only avoid introducing high frequency noise to fusion image,but also improve the visual effect of the fusion image effectively.In this paper, four typical image fusion methods were used to compared with this algorithm. Compared with the former three methods and the fourth method, we conclude that NSCT is applied to fuse images can obtain the higher quantity fusion image;comparison of the fourth method and the fifth method can indicate that this fusion rule is more efficient. The fusion images of simulation are evaluated through subjective observation and objective indexs.Subjective observation is qualitative judgment and direct sensation on fusion images,five objective indexes is quantitative analysis on fusion images. From the result of subjective and objective evaluation,this image fusion method is better than tradition methods.