|
With different images of redundancy, complementarity, new image of better quality, more information can be obtained by the image fusion. As an important aspect of data fusion, image fusion with its intuitive nature of the data, visualize, win a wide range of attention and study.Image fusion is divided into three part:the pixel-level, the feature-level, and dec is ion-lev el, in this paper, the rules for multi-sensor image fusion research based on the pixel level, first introduced the basic principle of image fusion, including image sensors, integrated system architecture, the integration method used, respectively, from the perspective of quantitative and qualitative evaluation of the quality of fused images. A comprehensive analysis has been done on the wavelet using the lifting-scheme, and multi-scale, and multi-resolution conversion tool. Focused on the spatial domain image fusion methods and transform domain methods, and image fusion rules, a fusion rule is proposed which can be adaptive to adjust the parameters of the fusion process based on the lifting-scheme wavelet, eliminating the influence of the objective factors such as light to the image fusion, and improve the integration process automation. In the human body’s medical images, X-ray CT image and MR image are acquired from different sensors, the subjective performance and objective performance of the processing image have been greatly improved.OpenCV is an excellent open-source computer image library, mainly using C/C++ language, which can run on Windows/Linux platform. A image fusion program based on the MFC and OpenCV was projected and finished.
|