Research On Multisensor Image Fusion Algorithms

The development of remote sensing and sensor technology has brought the abundance of data source, the data coming from diverse sensors have different time, space and spectrum resolution. They may also have different polarization. This produces the need that the processing techniques can effectively combine information from different sources into a single composite for interpretation. That is multisource information fusion. In image-based application fields, image fusion, as an important branch of it, has emerged as a promising research area. It has abroad applications in defense system, geoscience, medicine image processing, robot vision etc. Image fusion can integrate different images coming from multiple sensors as one image in order to obtain fusion result which has more reliability, less ambiguity and better understanding. After fusion, the result is more appropriate for human vision and computer processing, such as detection, classification, recognition and understanding.This dissertation mainly aims at the research of multisensor image fusion algorithms in pixel level. Firstly based on the review of the former research of fusion algorithms, the wavelet-based image fusion algorithms are investigated in focus and a wavelet-based image fusion algorithm using gradient selective rule is proposed. Secondly, according to the inherent characteristics of remote sensing images, two pertinent algorithms are proposed, one is the region-based remote sensing image fusion method, the other is the image fusion of SAR(Synthetic Aperture Radar) and optical image based on complementary information characteristics. Finally, the theory of ridgelet and curvelet which are proposed as a new multiscale analysis method in recent years is investigated and ridgelet and curvelet transform based image fusion are propsed. Remote sensing images are mainly used in this dissertation. SAR and optical images are in focus.The main contributions of this dissertation are summarized as follows:1. Based on summarizing the wavelet-based image fusion methods at present, it's current research and future trend are analyzed from tow respects: the form of wavelet decomposition and fusion rules. SAR and optical image fusion is performed using wavelet transforms with different decomposition schemes and wavelet basis. The characteristics of different wavelet transforms are investigated.Considering the conducted experiment and computational complexity, recommendations on choosing a particular wavelet transform to produce satisfactory fusion result in SAR and optical image fusion in different applications are given. The conclusion is also a reference for other kinds of remote sensing image fusion based on wavelet transform.2. A wavelet-based image fusion algorithm using gradient selective rule is proposed. The computation of local average gradient and global gradient is...