| Digital image processing is the premise and key step of image segmentation, image recognition and feature extraction, so it is the research focus and important content in the field of image processing. There has been a lot of kind of digital image edge detection algorithm, but most of them still have shortcomings, such as losing effectiveness in certain noise or environment, intermittent of edges, and so on. In this paper, it mainly puts focus on the edge detection in images affected by noise, and Fractional Fourier Transform(FRFT), as a booming tool in the field of time-frequency analysis, is applied to image edge detection. FRFT has not only excellent characteristics of Fourier transform, but also the time-frequency characteristics. So, the FRFT can also be applied in image processing.Aiming at the research of digital image edge detection, the main content of this thesis is described as follows:1. It describes the basic concept of the image edge detection, then summarizes some classical edge detection algorithms and some new edge detection algorithms, Finally, the common evaluation criterion of edge detection is given. The advantages and disadvantages of these different algorithms are analyzed through the experiments.2. 1-D and 2-D fractional Fourier transform are detailed described, and their characters are also elaborated in detail. The concept of frequency band decomposion is applied in the FRFT, and the structure of decomposion in the domain of FRFT is given. Experiment shows the properties of this decomposition, image can be reconstructed by using the low frequency subband.3. By combing the traditional Sobel operator and the low-frequency subband image after the decomposion in the domain of FRFT, edges could be extracted. Besides, multiscale decomposition and edge tracking algorithm are also taken into consideration. Experiment results show that the accuracy of edge detection is improved, and a much better edge detection effect is obtained. |
PDF Full Text Request