基于自适应矢量参数的彩色图像梯度计算方法

A New Method for Calculating Color Image Gradient with Adaptive Vector Parameters

由于灰度图像梯度计算方法难以直接推广到彩色图像梯度计算中,提出了一种基于自适应矢量参数的彩色图像梯度计算方法。该方法在HSV彩色空间内利用空间矢量求导方法计算彩色图像梯度,并结合彩色图像直方图均方误差得到HSV三分量的矢量参数,从而实现基于自适应矢量参数的彩色图像梯度计算方法。该方法解决了传统彩色图像梯度计算方法难以获取图像的细节梯度信息的问题,以及现有的彩色图像梯度计算方法不够稳定,普适性差的问题。实验结果表明,针对多种不同类型的彩色图像,该方法均能得到较好的梯度图像,较传统的方法获得更多的图像细节。

To our knowledge, it is difficult to directly extend the grayscale image gradient calculation method to color image processing. Hence, we propose the color image gradient calculation method mentioned in the title. Its core consists of： （1） we calculate the color image gradient with the spatial vector derivative in the hue, saturation, val- ue （HSV） color space ; （2） e use the ratio difference of the three components of HSV in different types of image and the mean-square error of the color image histogram to determine the sizes of their three vector parameters and obtain a new gradient calculation formula as given in eq. （16）. Our new method overcomes the difficulties of the traditional color image gradient calculation methods in obtaining an image＇ s detailed gradient information and the low robustness and poor universality of the existing color image gradient calculation method. The simulation results, given in Figs. 6 and 7, and their analysis show preliminarily that our method can obtain rather good gradient color images and more detailed gradients than the traditional image gradient calculation methods, thus being suitable for various types of color image.