基于参数最优控制的并行阵列稳态系统的彩色图像信号增强

Color Image Signal Enhancement in Parallel Array Steady-State System Based on Parameter Optimal Control

摘要: 本文针对均匀退化场景下的彩色图像信号，以增强彩色图像对比度与抑制有害噪声为目的，分析BPAM信号、灰度图像与彩色图像实现信号增强的联系与区别，推导了基于并行阵列双稳态系统的最优稳态信噪比参数控制，提出了基于物理先验的光照映射估计与并行阵列随机共振系统的双尺度控制算法(Stochastic Resonance in Illumination Map Estimation, SR-IME)。首先结合物理背景的先验估计彩色图像的光照通道，分别进行光照特征与数字特征的一维编码。其次将信号调制为BPAM信号，利用随机共振系统实现信号增强。从系统参数分配机制出发，将最优控制下的输出信号解调，反演得到增强后的彩色图像信号。结合主观评价和客观评价指标，与直方图均衡化HE、单尺度Retinex (SSR)，多尺度Retinex (MSR)和基于HSV分解的双稳态单系统进行了比较，结果表明SR-IME视觉效果最佳，信息熵为7.82，相比于HSV通道分解的单系统随机共振提高了5%，噪声方差与NIQE指标有明显改善。实验结果表明，在参数最优控制下的并行阵列稳态系统能有效实现彩色图像信号增强。

Abstract: In this paper, aiming at color image signals in uniformly degraded scenes, in order to enhance the contrast of color images and suppress harmful noise, we analyze the relationship and differences between BPAM signal, gray image and color image to achieve signal enhancement, derive the optimal steady-state signal-to-noise ratio parameter control based on parallel array bistable systems, and propose a dual-scale control algorithm based on physical priors for Illumination Map Estimation (SR-IME) for parallel array stochastic resonance systems. First, the illumination channel of the color image is estimated based on the priori of the physical background, and one-dimensional encoding of illumination features and digital features is carried out respectively. Secondly, the signal is modulated into a BPAM signal, and a stochastic resonance system is used to achieve signal enhancement. Based on the system parameter allocation mechanism, the output signal under optimal control is demodulated and the enhanced color image signal is obtained by inversion. Combining subjective evaluation and objective evaluation indicators, compared with histogram equalization HE, single-scale Retinex (SSR), multi-scale Retinex (MSR) and a bistable single system based on HSV decomposition, the results show that SR-IME has the best visual effect, with an information entropy of 7.82, which is 5% higher than the single-system random resonance based on HSV channel decomposition, and the noise variance and NIQE indicators are significantly improved. Experimental results show that the parallel array steady-state system under optimal parameter control can effectively enhance color image signal.