Abstract:

In order to make the details of the grayscale image more prominent, the visibility is enhanced. This paper presents an image enhancement method based on discrete cosine transform and quantile algorithm. The low frequency component of the image is extracted by discrete cosine transform, while the high frequency component remains unchanged. The quantile of the low-frequency components is subdivided to make the enhancement level of the low-frequency components involved in the enhancement process selective, and then the sub-histograms are equalized by histogram to enhance the image contrast. The processed low frequency component and the untreated high frequency component are inversely converted to obtain the enhanced image. In this paper, Mongolian furniture patterns are selected, and compared with the traditional adaptive histogram equalization algorithm and directional adaptive interpolation algorithm, this method has a better visual effect in the enhancement of Mongolian patterns, and its evaluation index is also significantly improved.

Key words: quantile, image enhancement algorithm, high frequency component, low frequency component, discrete cosine transform

摘要：

为了使灰度图像的细节更加突出、可视性增强，提出一种基于离散余弦变换与分位数算法结合的图像增强方法。通过离散余弦变换提取出图像的低频分量，图像高频分量保持不变。对低频分量进行分位数的细分，使参与增强过程低频分量的增强级别具有选择性，再分别对这些子直方图进行直方图均衡化，使图像对比度增强。将处理后的低频分量与未处理的高频分量进行逆变换，得到增强后的图像。选取蒙古族家具纹样，与传统的自适应直方图均衡化算法及方向自适应插值算法相比较，提出的方法在蒙古族纹样增强方面具有更好的视觉效果，其评价指标也有显著提高。

关键词: 分位数, 图像增强算法, 高频分量, 低频分量, 离散余弦变换