Research On The Method Of Image Matching Based On Semi-Variance Function

Image matching is an important technique in auto-measurement of digital photogrammetry. The application of image matching is popular way in digital sampling, resampling, three-dimensional measurement, stereo modeling, multi-source image merging and other related fields. With the development in information acquisition technique, some new applications and demands come out, so the rigorous matching algorithm is demanded. Based on the analysis of traditional image matching methods and the introduction of semi-variogram function from statistics, this paper developed an image matching notion by the method of semi-variogram function, studied the match-element of semi-variogram function image matching, constructed the semi-variogram function image matching similarity measurement, designed and realized four image matching algorithm by semi-variogram function. Based on the testing data analysis, we firstly chose an appropriate pixel length of semi-variogram function image matching template, tested the feasibility of the four algorithm mentioned above. Then we made another test to analyze the performance such as matching accuracy, speed and, probability of semi-variogram function by contrasting the correlation coefficient matching based on gray (pixels). The results indicated that the effect of semi-variance neighborhood matching algorithm works well, which can maintain a good accuracy of correlation coefficient and shorten the time greatly. The performance of other three algorithms can be improved by the optimization strategy. The results provide a desirable solution towards the relationship between accuracy and run time, which furnishes a kind of bran-new thought and way for image matching.