基于局部梯度极值点的经验模式分解图像增强

Image Enhancement of Empirical Mode Decomposition Based on Local Gradient Extreme Point

为解决传统2维经验模式分解获取图像细节能力不足的问题,提出一种基于局部梯度极值点的改进BEMD图像增强方法。根据梯度对图像细节信息的强挖掘能力,基于像素点4个2维方向上的极值条件来寻找图像的局部极值点,对图像进行经验模式分解并确定内蕴模式函数,结合大尺度梯度保留、小尺度梯度去除的思路,达到在图像增强的同时又抑制噪声的目的。实验结果表明:与传统的图像增强算法相比,该方法具有更强的细节捕捉能力。

Aiming at the inadequacy of traditional 2 D empirical mode decomposition in acquiring image details, the improved BEMD image enhancement algorithm based on local gradient extreme point is proposed. Based on the strong ability of gradient to mine image detail information, according to four 2 D extreme conditions of pixel point to find out image local extreme point, carry out empirical mode decomposition of image and ascertain connotation mode function, combine with large-scale gradient preservation and small-scale gradient removal, achieve the purpose of image enhancement and noise suppression. The experiment results show that the proposed method has better detail capture ability than the traditional image enhancement algorithm.