摘要
针对传统边缘检测方法(gauss＿laplace、kirsch、robert等)的不足,提出了基于HVS(humanvisualsystem)函数的边缘检测方法.根据HVS函数的特点,对频域中HVS函数进行反傅立叶变换,将空间域中该函数的部分函数值作为特定的模板参数,并与原图像进行卷积来实现边缘检测.对比性实验结果表明,用该方法检测边缘精确度高,效果很好,尤其丢失的边缘细节信息很少.
Based on the analysis of the shortages of traditional edge detection of image, such as Gauss-laplace method, Kirsch method and Robert method, a new edge detection based on HVS(human visual system) function is proposed. The HVS function, HVS function is inversely Fourier transformed in frequency domain, and the method is realized by calculating some value of the function in spacial domain as convolve nucleus. Lots of comparative experiments had improved that accuracy and whole effects of edge detection of image are higher than the traditional methods. The applied scope can be extremely extended. All sorts of detail information of edge are almost not lost.
出处
《西安工业学院学报》
2003年第4期330-333,共4页
Journal of Xi'an Institute of Technology
