Studies On Optimal Matching Technology And Algorithm Of Multi-sensor

The multi-sensor image matching technology is the processing of matching two ormultiple images obtained from different sensors, different perspectives or at differenttimes, and this technology is a fundamental issue in image processing. This paperfocuses on the primary field of the optimal alignment of multi-sensor ---time matching,space matching, image matching, and has done an in-depth research.Space match and time match is a match for the sensor to correct itself, which is apreliminary job that has to be done ahead in the practical application for imagematching and integration. As for space registration, the methods, such as maximumlikelihood, UKF (unscented Kalman filter), information entropy, are often used toachieve registration. After finishing collecting and sorting the works of our predecessors,I had done a scientific and systematic explanation and research in this field. As a result,I mainly recommend the method of geometric correction, and a desired result can be gotbyusing this method for rapid multi-sensor registration.Time registration is to synchronize at the same time the unsynchronizedmeasurement information from various sensors based on the same goal. On the basis ofsumming up in the predecessors'algorithm, I focused on the study of adaptive trackingalgorithm, which has more precise registration and speed proved bymany experimentsFeature-based image registration method is one of the most commonly usedmethods of image matching ,and its greatest strength is its ability to transform thevariety of analysis on the entire image into the characteristics of the image(characteristics, characteristics of the curve, etc.), Thus it greatly reduces the computingcapacity in image processing. The accuracy of feature extraction and the positioningprecision have a great impact on the whole process of registration. Based on the analysison existing feature extraction methods, the article focuses on the application of the edgecharacteristics in the field of the scene matching, and points out the method to improvethe feature extraction based on some Hausdorff Distance.Finally, I do a summarization of this paper, and look forward to continuing thein-depth studyin the field of the optimal registration of multi-sensor.