Image Edge Detection Based on Wavelet-Space Entropy and Morphology
The edge detection of the image with strong noise is a difficulty and hot-point in the field of image recognition. An algorithm of edges detection based on wavelet-space entropy and morphology was presented to achieve the sub-pixels edge of the strong noise image, according to the trait that the wavelet-space entropy could directly reflect the weights of the sub-image edges. The image was decomposed by wavelet. And the series of sub-images edges were detected respectively using the multiple structure elements and the multi-direction elements. The weights were determined adaptively by calculating the waveletspace entropies of the edges detected in the subimages. The edges disposed by the forth two methods were fused to achieve the final image edges. In the end, compared with some conventional edge detection methods, the experiment was made. Results showed that the proposed algorithm could adaptively detect more details and more complete and successive edges. It had stronger ability to depress the noise and locate the edge. And the sub-pixel edges detection of the strong noise image could be achieved easily. The adaptability and real-time performance was better than conventional ones.
edge detection wavelet-space entropy multiple structural and multi-scaleadaptive mathematical morphology
HOU Hong-Hua GUI Zhi-Guo
The Key Laboratory of the State Education Ministry on Instrumentation Science & Dynamic Measurement, The Key Laboratory of the State Education Ministry on Instrumentation Science & Dynamic Measurement,
国际会议
太原
英文
438-442
2011-02-26（万方平台首次上网日期，不代表论文的发表时间）