基于金字塔模型和各向异性滤波的分层自适应图像增强算法

结合拉普拉斯金字塔模型分解和前后双向异性扩散算法,提出一种分层自适应图像增强算法.该算法首先进行图像的高动态范围压缩,然后采用拉普拉斯金字塔模型方法将原始图像分解为不同尺度和频率下的带通图像序列.根据不同频率层图像的纹理方向特征,设计自适应参数法修改扩散传导方程的参数,在不同频率图像层上分别实现噪声平滑和边缘特征的增强.仿真实验通过与其它图像增强算法进行比较,评价结果表明,提出的分层自适应图像增强算法的处理效果良好,定量评价指标大幅改善.

Layered Adaptive Image Enhancement Based on Laplacian Pyramid and Anisotropic Filter

An improved layered adaptive image enhancement method is proposed by combining the Laplacian pyramid mode and anisotropic filtering. First high dynamic range compression was realized by bilateral filter. Then Laplacian pyramid model was adopted to decompose the image into a set of band pass filtered images. Moreover, an improved forward and backward anisotropic diffusion filtering algorithm (FBAD) was designed, which can adjust the diffusion parameters adaptively. Applying the FBAD filtering algorithm to layered band pass filtered images, through selecting different diffusing thresholds, the noise of image could be removed and the edges of the images be sharpened. Comparative study on experimental results of FBAD and other image enhancement algorithms shows the effectiveness of the proposed method.