基于自适应伪轮廓消除滤波的比特深度增强

Adaptive False Contour Elimination Filter-Based Bit Depth Enhancement

  • 摘要: 比特深度增强具有巨大的应用价值,如低比特图像采集、高比特图像显示、图像压缩等。然而,现有的比特深度增强算法或者不能有效抑制低比特图像中存在的伪轮廓,或者需要高性能的计算单元和大量的训练样本,缺乏实用性。为此,提出了一种基于自适应伪轮廓消除滤波的比特深度增强算法:FACE-BDE。FACE-BDE由3个模块组成:伪轮廓消除滤波器设计模块、自适应滤波器大小选择模块和伪轮廓区域检测模块。实验表明,所提算法能在消除不同大小的伪轮廓时保留真实轮廓和细节,使其在测试集上取得了0.28 dB的增益,伪轮廓的可见性也明显弱于同类算法。

     

    Abstract: Bit depth enhancement has great application value, such as low-bit image acquisition, high-bit display, and image compression. However, existing bit depth enhancement algorithms either cannot effectively suppress false contours that appear in low-bit images or require high-performance graphical processing units and a vast number of training samples, making them lack practicality. To this end, we propose an adaptive false contour elimination filter-based bit depth enhancement algorithm, named FACE-BDE. FACE-BDE consists of three innovative modules: a false contour elimination filter design module, an adaptive filter size selection module, and a false contour region detection module. Experiments show that our algorithm can preserve real contours and details when eliminating false contours of different sizes, thus achieving 0.28 dB gain on the testsets, and the visibility of false contours is significantly unnoticeable than that of similar algorithms.

     

/

返回文章
返回