基于SVR的图像增强方法

Image Enhancement Based on SVR

传统的图像增强方法是采用直方图均衡化来处理图像,利用这种方法处理后的图像,虽然使图像的灰度等级得到拉伸,但是却使图像变得过亮,部分细节消失。为改善图像增强的效果,利用支持向量回归原理(SVR)的方法来处理图像,试验取得了很好的结果。试验结果表明,采用SVR的图像处理方法来增强图像,可以很好地保持图像原有的灰度特征,突出图像的细节,并且解决了传统的直方图均衡化方法使图像过亮和部分细节丢失的问题。

The traditional method for image enhancement is histogram equalization. Using this approach to the image, although it enables the gray levels to obtain the stretch, but actually causes the image becomes excessively bright, partial detail vanishing. To improve the effectiveness of image enhancement, use support vector regression （SVR） principle approach to handle images, the experiment has obtained a very good result. The test result indicated that using SVR principle to enhance the image, can maintain a good image of the original gray feature, the prominent picture detail,and avoid the traditional histogram equalizing method causing the image to be excessively bright and the partial details lost.