低照度图像增强的自适应同态滤波算法研究
Research on Adaptive Homomorphic Filtering Algorithm for Low Illumination Image Enhancement

针对传统同态滤波计算复杂、实时性差、参数多且最优参数获取困难等问题，本文提出一种基于自适应同态滤波的低照度彩色图像增强算法。为避免增强图像颜色失真，基于HSV色彩空间，仅对V分量进行增强处理；结合离散余弦变换(Discrete Cosine Transform, DCT)及指对变换实现图像空间域与频率域的相互转换；基于sigmoid函数构造单参数同态滤波函数，以峰值信噪比(PSNR)和结构相似性(SSIM)为指标，建立一种图像评价模型，并采取等步长间隔法实现最优参数的自适应选取；选取信息熵、平均梯度和对比度对道路及建筑图像进行定量评价，实验结果表明，本文算法可有效校正低照度图像亮度不均问题，提高了图像对比度，图像细节信息得到了进一步丰富。
Aiming at the problems of traditional homomorphic filtering, such as complex computation, poor real-time performance, many parameters and difficulty in obtaining optimal parameters, this paper proposes a low illumination color image enhancement algorithm based on adaptive homomorphic filtering. In order to avoid color distortion of enhanced image, only V component is enhanced based on HSV color space. Combined with discrete cosine transform, exponential transformation and loga-rithmic transformation, the interconversion of image space domain and frequency domain is realized. A single parameter homomorphic filtering function was constructed based on sigmoid function. Taking peak signal to noise ratio (PSNR) and structural similarity (SSIM) as indexes, an image evaluation model was established, and the equal step interval method was adopted to realize the adaptive selection of optimal parameters. Information entropy, average gradient and contrast were selected for quantitative evaluation of road and building images. The experimental results show that the proposed algorithm can effectively correct the uneven brightness of low illuminance images, improve the image contrast and enrich the image details.