Research Of Information Fusion And Enhancement Of Multi-Source Image

Multi source image information fusion is an important image processing method that synthesizes two or more images from different sensors into a new fusion image through a special algorithm. In this way it not only can make the new fusion image contain the complementary information from every single-source image of single source image, expand the sensing range of space-time and improve the accuracy and robustness of observation, but also will make the image with higher reliability, better intelligibility and easier access to meeting specific needs. Therefore the technology of multi-source image fusion has been widely applied in the military, remote sensing, civilian, medical science and other fields.In this paper, the theory of image fusion, image fusion method and image quality evaluation method have been introduced firstly. And then based on the features of infrared images and visible images, a new improved registration method has been put forward, which takes normalized mutual information as the similarity measure of heterologous image. In order to speed up the algorithm, the pyramid decomposition method has been used to realize algorithm optimization. Registration experiments with the infrared and visible light images have proved the validity of the algorithm.Based on the complementary relationship between the information of heterogenous images, the wavelet transform fusion method that aims at the target property sphere has been proposed. Firstly, extraction of infrared image target characteristic region has been realized by using the fuzzy C-means clustering (FCM) method for local information fusion of heterogenous image target characteristic and the effect that infrared target information to add to the visible light image has been achieved. Secondly, the final fusion image has been got by using wavelet transform method to realize fusion of infrared images and the secondary fusion image from the first step. The experimental results of the algorithm proposed in the paper and other fusion algorithms have been analyzed and compared based the image quality evaluation standard.In order to improve the fusion image identification, the method of information enhancement has been studied in this paper for multisource fusion image. Based on multiscale smoothness a new enhancing algorithm has been presented and the image information enhancement can be achieved by enhancing the detailed difference between the original image and Gaussian smooth image in different scales. The effect of this method is also verified through experiments in this paper.At last, the multi-image-source imaging platform has set up with Guide IR256 cooling MW thermal imager and Daheng DH-HV1302UC industrial camera, and the control circuit board applied to imaging platform has also been designed. The software applied to multiple image source acquisition, control and image fusion processing has been developed under the VC++6.0 programming environment and it can realize Quasi-real-time information fusion of the infrared image and visible light image.