Research And Application On The Key Techniques Of Multi-Source Image Fusion

In recent years, image fusion has been an important and useful new technology in the fields of image understanding and computer vision. Because of the limitation of single sensor by its own performance, single sensor only works in certain conditions and within certain scope, so the data provided by single sensor is limited. However, several source image from multiple sensors include the complementary and redundant information, multi-sensor image fusion technology can synthesis of complementary information and remove redundant information between images, to obtain more accurate 、 more reliable and more comprehensive information.The algorithms of multi-sensor image fusion aiming at pixel level are studied in this paper. The unified framework based on the principle of multi-scale is established for the multisource image filtering、registration and fusion. The framework of the whole process of the multi-sensor image fusion system is given in this paper. In this paper, the main work is as follows:1. Firstly, the research background and meaning of multi-sensor image fusion are expounded. Secondly, the applications and the classification of image fusion are introduced. Finally, the develop and research status of image fusion are summarized.2. In this paper, the denoising algorithm directing at images from different sensors is studied and multi-scale morphology denoising algorithm with multiple structural elements is used, effectively removing noise and well protect the edges of image. The simulation analysis from several sets of image data and the comparisons with various denoising algorithms verify the effectiveness of the algorithm in this paper.3. In this paper, the multi-sensor image registration algorithm is deeply discussed and the image edge feature is used to registration. For different edge detection algorithm, the simulation analysis and comparison with several sets of image data verify the effectiveness of the multi-scale morphological edge detection operator using several structural elements. The algorithm in this paper can attain more complete and more continuous image edge.The cross-correlation information is used to registration as a similarity criterion to attain space coordinate transformation parameters, with the edge of image. The image is registration using the parameters. Using image edge as registration feature overcomes the defect of the time of registration too long.4. Multi-sensor image fusion based on multi-scale decomposition is used in this paper and a variety of fusion algorithms and fusion rules directing at image decomposition coefficient are studied. In the process of fusion, the fusion rules based on regional characteristics of decomposition coefficient is adopted, considering not only the low frequency information to clearly describe the target contour, but also combine with the high frequency to get the detail of the image. Using the combination method of subjective and objective foe the evaluation of the fused images, the simulation analysis and comparison with several sets of image data verify the algorithm in high application.5.The software system applied in the fusion of visible image and infrared image is developed in this paper integrated the above method. The software system can realize the functions of image reading, image denoising, image registration, image fusion, image fusion effect evaluation and data storage and remove.