As an important branch of multisensory data amalgamation, image amalgamationis widely used in areas such as military, network security and medical imagery. Recently,with the fast development of infrared and low light level sensor, the amalgamation ofinfrared and low light level images becomes a hot topic in the field of night vision.Discovering and observing the opposite earlier, further and better under the weak lightcondition has become a deciding factor in the modern war. The advantages of the imageoutput of the infrared thermal imager are as follows: First, the contrast of the image isrelatively large; Second, the image is seldom affected by the environment. However, thequality of the observed image is poor. The image outputs by low light level night visiondevice are of high quality and resolution, but it has following shortcomings: the layer ofimage is not distinct and environmental impact is obvious. Therefore, if these twoimages are amalgamated with certain algorithm, the observation capabilities of the nightvision systems will be greatly improvedThe dissertation analyses the image source and image characteristics of the infraredimages and low light level image and presents some methods in image preprocessing,image amalgamation algorithm and the amalgamation image assessment, whichprovides theoretical guidance and technical support to image amalgamation system. Theinnovated aspects are:1. By using image processing board, TMS320DM642digital signal processor as itscore, to carry out the amalgamation processing.2. To select the appropriate image fusion algorithms and applies it to the fusionsystem by using MATLAB simulation.3. The images here enhanced and the corresponding match has been done duringthe pretreatment process. Furthermore, by using the neural network algorithm and theDSP program, the amalgamation program runs.4. By experiment amalgamation images have reached a real-time level whichachieves a satisfactory result. |