基于VR学习环境的色觉异常学习者感知补偿研究

【作者】 张宁；

【导师】 刘振；

【作者基本信息】 曲阜师范大学 ， 教育技术学， 2025， 硕士

【摘要】 色觉异常(色盲/色弱)是一种由于视觉系统功能异常导致的色彩认知障碍,全球约2亿人受此影响,其中男性患病率约5%-8%。色觉异常患者在红、绿等颜色的分辨上存在显著困难,这不仅影响他们的日常生活和职业发展,也给教育学习带来较大挑战。特别是在色彩识别和学习材料的处理上,色觉异常学生常面临障碍,导致其学习效率降低,成绩受到影响。然而,现有医学色觉障碍矫正技术普遍存在潜在视觉损伤风险,而数字视觉辅助技术可通过图像处理手段提升色觉异常学习者在易混淆颜色上的对比度。因此,亟需开发非侵入性的数字化辅助解决方案,提升色觉异常学习者在教育环境中的色彩感知能力。基于此,本研究通过深度学习技术,提出了一个色觉异常学习者的感知补偿模型,并在VR学习环境中进行验证。具体研究内容如下:针对色觉异常群体感知图像数据集缺失的问题,本研究通过模拟不同类型色觉障碍的视觉感知特征,利用Daltonize差分算法实现RGB-LMS颜色空间转换,生成适用于不同色觉障碍类型的矫正图像集。为提高数据集多样性,采用数据增强模块进行几何变换、颜色扰动等预处理操作,以增强后续模型的泛化能力。通过此方法,扩充了包含1928个样本的图像数据集,有效支撑了模型的训练。针对现有色彩补偿模型在保留自然度方面的不足,本研究提出了一种基于pix2pix 网络的色彩感知补偿模型。该模型结合了 ECA注意力机制和多空间约束机制,优化了对图像色彩属性和空间特征的提取能力,同时实现了不同类型色盲和色弱学习者的色彩感知补偿效果。实验结果显示,所提模型在提升色彩辨识度的同时较大程度地保留了图像的自然属性,补偿效果显著。具体而言,在色觉异常的不同视角下,补偿图像与原图的SSIM值均在0.85以上,其颜色丰富度也得到了不同程度的提升,数据表明该模型能够有效实现不同类型色盲与色弱的色觉感知补偿,实现更符合人眼视觉特性的补偿效果。针对色觉感知补偿模型在VR学习环境中开发及效果评估,本研究开发了一个高度沉浸的VR学习环境,结合色彩感知补偿模型,实现自然图像、石原测试图和学习图像的色彩转换与呈现。通过主客观相结合的评估方法,从颜色区分度、信息解析率和视觉舒适度三个维度,评估了模型在VR学习环境中的效果。实验结果表明,本文提出的色觉异常感知补偿模型能够显著改善色觉异常学习者对图像内容的感知效果,同时提供了更舒适的视觉体验,为他们在VR学习环境中的有效参与创造了条件。更多还原

【Abstract】 Color vision deficiency(color blindness/color weakness)is a perceptual disorder caused by abnormalities in the visual system,affecting approximately 200 million people worldwide,with a prevalence of 5%-8%in males.Individuals with color vision deficiency experience significant difficulty distinguishing colors such as red and green,which not only impacts their daily life and career development but also presents substantial challenges in educational settings.In particular,these individuals face barriers in color recognition and processing learning materials,leading to reduced learning efficiency and performance.However,existing medical corrective technologies for color vision deficiencies often carry potential risks of visual damage.In contrast,digital visual assistive technologies can enhance contrast in colors that are commonly confused,thereby improving color perception in individuals with color vision deficiency.Therefore,there is an urgent need to develop non-invasive digital assistive solutions that enhance the color perception abilities of learners with color vision deficiencies in educational environments.Based on this need,this study proposes a perceptual compensation model for learners with color vision deficiencies using deep learning techniques,and the model’s effectiveness is validated in a virtual reality(VR)learning environment.The specific research content is as follows:To address the issue of a lack of perceptual image datasets for the color vision deficiency population,this study simulates the visual perception characteristics of various types of color vision disorders,utilizing the Daltonize differential algorithm to achieve RGB-LMS color space conversion,thereby generating corrected image sets suitable for different types of color vision deficiencies.To enhance dataset diversity,data augmentation techniques such as geometric transformations and color perturbations are employed as preprocessing steps,thereby improving the generalization ability of subsequent models.Through this approach,a dataset containing 1,928 samples was constructed,providing effective support for model training.In response to the limitations of existing color compensation models in preserving naturalness,this study proposes a color perception compensation model based on the pix2pix network.The model incorporates the ECA attention mechanism and multi-space constraint mechanism to optimize the extraction of color attributes and spatial features from images,achieving color perception compensation for learners with various types of color blindness and color weakness.Experimental results show that the proposed model significantly enhances color recognition while preserving the naturalness of the images.Specifically,the SSIM values between the compensated images and original images are above 0.85 for various types of color vision deficiencies,with the color richness of the images also being significantly improved.These data demonstrate that the model effectively compensates for the color perception of different types of color blindness and color weakness,providing a compensation effect that aligns more closely with human visual characteristics.Regarding the development and effect evaluation of the color perception compensation model in a VR learning environment,this study developed an immersive VR learning environment integrated with the color perception compensation model,enabling the color conversion and presentation of natural images,Ishihara test images,and learning images.The effectiveness of the model in the VR environment was assessed through a combination of subjective and objective evaluation methods,focusing on color distinguishability,information parsing rate,and visual comfort.The experimental results show that the proposed color vision deficiency compensation model significantly improves the perceptual effectiveness of learners with color vision deficiencies and provides a more comfortable visual experience,facilitating their effective participation in the VR learning environment.更多还原