The Research Of Image Fusion Technology Based On The Nonsubsampled Contourlet Transform

With the image sensor technology, multi-sensor information fusion technology as an emerging area of research, has greatly improved the lot of industrial measurement systems and intelligent control system performance. Image fusion as an effective integration of information technology has been widely used, such as in the machine vision, military detection, identification and medical imaging lesions in areas earth remote sensing. The main research content of this article are based on Nonsubsampled Contourlet transform image fusion technology and its application.The basic concepts and theories of the image fusion technology are analyzed and summarized in this thesis. On the basis of the current image fusion algorithm, this thesis has analyzed and researched some typical image fusion algorithms, and Based on the properties of NonSubsampled Contourlet Transform, shift invariance, multiscale and multidirectional expansion, an image fusion algorithm based on the Nonsubsampled Contourlet Transform (NSCT) is proposed. The main thesis work and innovation are as follows:1. For wavelet transform has many defects and normal transform an ordinary mixer Contourlet transform, selection the Nonsubsampled Contourlet transform as source image frequency domain transform, be able to fully exploit the question of the direction of integration of image information, for the opposite sex based on High-dimensional express more sparse, reflects the multi-scale, multi-directional; and because of Nonsubsampled Contourlet transform invariance and effective to avoid the emergence of the Gibbs phenomenon.2. In the design of fusion rule on questions of Nonsubsampled Contourlet transform decomposition obtained coefficient of low-frequency subband and high frequency subband coefficients are different design options: selection at low-frequency sub-band coefficients, a coefficient based on the definition Choice programs; at select high-frequency sub-band coefficient, the introduction of Pulse-Coupled Neural Network (PCNN) proposed an adaptive coefficient PCNN options.3. At the use of PCNN, using images corresponding to the mean as a PCNN neuron connection strengthβ, after PCNN fusion ignition judged to be a clear part of the image, and select a clear integration of image pixels as a high-frequency pixel, making PCNN with Adaptive must sexual. 4. Analysis and Comparison of the current variety of image fusion algorithms, and with the algorithm for horizontal / vertical contrast test, and implementation of multi-sensor image fusion, and fusion with good effect and the larger study of the meaning and value.Above-mentioned study, the initial settlement of the car in safe driving auxiliary lighting, weather and other environmental factors affect the driver on the road environment to determine the exact problem, and strengthen the visual image information for the driver to provide a reliable, safe vision information, and for the follow-up target identification, tracking, detection and provide a richer, moreaccurate information.