Research On Edge Detection And Corner Detection Algorithm Based On Adaptive Fractional Differential

The advantages of the theory in the field of digital signal processing are more andmore realized by people. Fractional differential applied to the image processing is anascent research topic. This paper has carried out some tentative exploration offractional differential theory applied to edge detection, corner detection, and have madesome research achievements. My research contents are rendered below:First, when the fractional differential applied to the image processing, the best orderof fractional differential needs to be specified by the researchers, affecting the actualapplication of the fractional differential. This paper choice the fractal dimension, whichcould express the complexity of the detail of the texture, as a parameter adaptive todetermine the order of differential, and this approach makes the fractional differential beapplied in occasions with high real-time such as video target tracing and video imagestabilization. However, the result of the calculation method of the fractal box dimensioncommonly used is very rough. This paper analyzes the disadvantage of the algorithmand proposes the improved method. Experiments show that the Adaptive fractionaldifferential based on improved fractal dimension has a better result than Integer-orderdifferential in the edge extraction of the image.Second, There are plenty of false corners emerged in the corner detection of theimage with high texture complexity by Harris algorithm, This paper analyzes the reasonthat causes the false corners and replace the integral order in the algorithm withfractional order to operate differential coefficient can improve the algorithm. In addition,because corner detection is a part of the applications such as image mosaic, Adaptivedetermination of the order of the fractional differential is necessary. In order not toaffect the computational efficiency, this article uses the original fractal dimensioncalculation methods and linear map the calculation results on the interval. TheExperiments show that the modified algorithm has wider applicability and higherprecision in the corner detection. Finally, the paper made an image stitchingapplications, Based on Improved Harris Algorithm.Third, this paper has developed an image stitching application, which is based onHarris corner detection algorithm. This application show that the corner could be usedto feather matching. In addition, the application interface is simple, little changes canintegrate more functions. This paper has a brief summary of the Researching status of the fractional differential applied to image processing. The paper also detailed analysisof the reasons for the fractional differential template to extract the texture has advantagewhen compared to the integer-order differential. This paper detailed introduces the threedefinitions of the fractional differential, the knowledge of the traditional edge detectionoperator, the subsequent processing of the edge detection, corner detection commonlyused algorithm, as well as three methods of calculating the fractal dimension.