【摘要】 针对鲁哥氏液固定导致浮游动物显微图像（如桡足类、枝角类及轮虫）对比度降低、颜色失真以及附肢等关键形态特征模糊的问题，提出了一种基于等效光学密度重构与Lab色彩空间的图像增强方法。该方法在亮度通道，基于朗伯-比尔定律构建浮游动物显微图像的等效光学密度图，并结合Gamma变换与自适应直方图增强重构亮度通道；在色度通道，构建基于局部方差的自适应加权模型，以抑制非均匀染色引起的色度偏差。对多组显微图像的对比实验表明：所提方法在平均梯度（AG）、对比度改进指数（CII）及增强测量评价（EME）等客观指标上均优于主流方法。以剑水蚤（Cyclops）测试集为例，AG与EME均值分别提升至22.68和3.60，所提方法有效提升了鲁哥氏液固定后浮游动物显微图像的视觉质量与特征辨识度，且运行效率满足自动化监测系统的实时性需求，为浮游动物种属的形态学鉴定提供了可靠的技术支持。更多还原

【Abstract】 Objective Zooplankton are crucial indicator organisms for aquatic ecosystems, and their community distribution is a key basis for evaluating water health. In practice, Lugol’s solution is commonly used to fix zooplankton samples for long-term preservation. However, the strong penetration and staining effect of Lugol’s solution causes the zooplankton bodies to appear yellowish-brown, leading to blurred morphological features such as limbs and antennae, which significantly increases the difficulty of taxonomic identification based on computer vision. Therefore, it is of great value to develop an image enhancement method specifically for Lugolfixed zooplankton to restore texture details and improve the accuracy of automated species identification.Methods In this paper, a zooplankton microscopic image enhancement method based on an optical density model and Lab color space is proposed. First, the original RGB image is converted into the Lab color space to separate the luminance(L) and chrominance(a, b) channels. In the luminance channel, an equivalent optical density(OD) map is constructed based on the Lambert-Beer law to characterize the contour and internal structures of the translucent zooplankton. Gamma correction and contrast limited adaptive histogram equalization(CLAHE) are then applied to the OD map to reconstruct the luminance channel and enhance local gradients. In the chrominance channels, an adaptive weighting model based on local variance and a hyperbolic tangent function is constructed to suppress the color deviation caused by non-uniform staining. The final enhanced image is obtained by fusing the reconstructed channels and transforming them back to the RGB color space.Results and Discussions Microscopic images of six zooplankton species, including Copepoda(Sinocalanus, Cyclops), Cladocera(Moina, Diaphanosoma), and Rotifera(Brachionus Calyciflorus, Brachionus Quadridentatus), are used as experimental objects. The performance of the proposed method is compared with multi-scale Retinex with color restoration(MSRCR), Dehaze, hematoxylineosin(HE)-based density separation, MLLP, and the deep learning-based zero-reference deep curve estimation(Zero-DCE) algorithms. Subjective visual comparisons demonstrate that the proposed method effectively sharpens the edges of limbs and purifies the background. In terms of objective quantitative evaluation, the average gradient(AG), contrast improvement index(CII), and measure of enhancement by entropy(EME) of the proposed method are significantly superior to those of the compared algorithms. For the Cyclops dataset, the AG and EME reach 22.68 and 3.60, respectively. The algorithm achieves a processing speed of 57.43 frames per second(FPS) for Cyclops, which is approximately 50% higher than the Dehaze algorithm, meeting the real-time requirements for automated monitoring systems.Conclusions To solve the problems of low contrast and color distortion in Lugol-fixed zooplankton images, a spatial-domain enhancement method combining a physical absorption model and adaptive chrominance correction is proposed. The experimental results show that the proposed method significantly improves the visual quality and feature identifiability of zooplankton microscopic images. By restoring key morphological features while maintaining high computational efficiency, this work provides effective technical support for the development of automated zooplankton identification and monitoring instruments.更多还原