IEEE Trans Nanobioscience. 2021 Jul;20(3):278-286. doi: 10.1109/TNB.2021.3064077. Epub 2021 Jun 30.
In this paper, a novel triple clipped histogram model-based fusion approach has been proposed to improve the basics features, brightness preservation and contrast of the medical images. This incorporates the features of the equalized image and input image together. In the initial step, the low-contrast medical image is equalized using the triple clipped dynamic histogram equalization technique for which the histogram of the input medical image is split into three sections on the basis of standard deviation with almost equal number of pixels. The clipping process of the histogram is performed on every histogram section and mapped to a new dynamic range using simple calculations. In the second step, the sub-histogram equalization process is performed separately. Approximation and detail coefficients of equalized and input images are separated using discrete wavelet transform (DWT). Thereafter, the approximation coefficients are modified using some basic calculation-based fusion which involves singular value decomposition (SVD) and its inverse. Detail coefficients are fused using spatial frequency features. This yields modified approximation and detail coefficients for an enhanced image. Finally, inverse discrete wavelet transform (IDWT) has been applied to the modified coefficients which result in an enhanced image with improved visual quality. These improvements are analyzed qualitatively and quantitatively.
本文提出了一种基于新型三截断直方图模型的融合方法，以提高医学图像的基本特征、亮度保持和对比度。该方法将均衡图像和输入图像的特征结合在一起。在初始步骤中，使用三截断动态直方图均衡化技术对低对比度医学图像进行均衡化，其中输入医学图像的直方图基于标准偏差分成三个部分，每个部分包含几乎相同数量的像素。对每个直方图部分进行裁剪过程，并使用简单的计算映射到新的动态范围。在第二步中，分别进行子直方图均衡化过程。使用离散小波变换（DWT）将均衡化图像和输入图像的近似系数和细节系数分开。然后，使用基于一些基本计算的融合来修改均衡化和输入图像的近似系数，其中包括奇异值分解（SVD）及其逆。细节系数使用空间频率特征进行融合。这为增强图像生成了修改后的近似系数和细节系数。最后，应用逆离散小波变换（IDWT）对修改后的系数进行处理，生成具有改进视觉质量的增强图像。这些改进在定性和定量方面进行了分析。
