Applied Research Of Image Edge Detection By 1?2 Order Fractional Differential

The image edge detection is the key step in the subsequent processing during image segmentation or pattern recognition. Edge line should not only conform to the actual observation from the macroscopic visual quality, but also be needed to have different evaluation method to the quantitative analysis. Many new operators are constantly emerging since image edge extraction technology is put forward. Scholars have deduced all kinds of new type detection operator from their respective areas.Because of the variety and complexity of the image, the precision of edge detection and the detection of the broken edge are the difficult points in the present study.Focusing on the issues of failing to pinpoint the edge information accurately and lacking texture detail of image by using integer order differential or 0?1-order fractional differential mask operators in digital image processing, a new 1?2-order edge detection operator based on Laplacian operator was proposed. Deduced from the definition of Riemann-Liouville(R-L), the 1?2-order fractional differential had the advantage in enhancing high-frequency signal and reinforcing medium frequency signal. In order to get more accurate edge segment, the algorithm combined the crack edge and the segment of the fractional order differential extraction by searching weak edges.The primary work of this paper includes:(1)Eight-direction 1?2 R-L order fractional order differential mask operator has deduced by fractional order calculus;(2)From the edge integrity?complexity?location accuracy and noise resistance to analyzed the result, the results demonstrate the validness of the proposed methods.(3)In order to skip over the discontinuity point with broken edge segment.According to the local information of pixels, a new method of hysteresis connection based on gradient features has proposed by R-L fractional order differential algorithm.