Optimizing Visual Communication in Online Classrooms Using Image Processing Technology

With the rapid development of information technology, online education has become an important component of the modern education system [1-3]. Especially during the pandemic, the widespread adoption of online classrooms has enabled more students to conveniently access high-quality educational resources [4]. Unlike traditional face-to-face teaching, the teaching format of online classrooms is influenced by various factors such as hardware devices and network environments. In particular, the presentation of the visuals has a direct impact on students' learning experience and learning outcomes [5-7]. To improve the quality of online classroom teaching, the application of image processing technology has increasingly become a key approach to enhancing visual communication effects.

In the teaching content of online classrooms, integrating excellent traditional culture has become an important direction for enhancing the cultural value and teaching depth of the class. Elements of traditional culture often contain rich historical connotations and artistic expressions. However, in online classrooms, how to effectively present these cultural connotations through visual communication methods relies on the support of image processing technology [8-11]. Through adaptive image adjustments and visual enhancement techniques, the quality of the classroom visuals can be improved, enabling students to focus more during the viewing process while deepening their understanding and memory of the teaching content, especially the traditional cultural elements [12-15]. Therefore, exploring how to optimize the visual effects of online classrooms through image processing technology has profound educational significance and cultural value.

Although some studies have already explored the issue of image optimization in online classrooms, most of them are limited to basic aspects such as image clarity and color adjustments, lacking adaptive adjustments of classroom visuals in diverse environments [16, 17]. Moreover, existing visual enhancement techniques mainly focus on improving video fluency and clarity, while research on how visual effects can align with teaching content, especially the precise optimization of cultural transmission, is still insufficient [18-20]. Therefore, how to enhance the visual communication effect of online classrooms through more refined and personalized technical means remains an urgent issue to be addressed.

The main research content of this paper includes two parts. On the one hand, through adaptive color correction technology, the color distortion problems caused by factors such as lighting and display devices in online classrooms are improved, making the colors of the visuals more vivid and natural, thereby enhancing the learners' visual experience. On the other hand, through visual enhancement technology, key content in the classroom is emphasized and reinforced, making it easier for students to grasp the main points and improving the interactivity and appeal of the classroom. Through these technical means, this paper aims to explore how to create more vivid and intuitive visual communication effects in the online classroom environment, thereby improving students' learning efficiency and teaching effectiveness.

The color deviation problem of online classroom images has been largely addressed through the compensation of color channels and white balance algorithm adjustments mentioned earlier. However, to further enhance the visual effect of the image and ensure that the details and layers of the visuals are clearer, this paper introduces a histogram stretching method based on Gaussian modeling. This method analyzes and adjusts the distribution of the image's histogram, effectively solving the issue of concentration in the histogram distribution, significantly optimizing the contrast and brightness of the image, and thus improving the visual communication effect of the online classroom visuals.

3.1 Gaussian modeling

By modeling the R, G, B channels of the online classroom image with a Gaussian model, finer color and brightness adjustments can be achieved, enhancing the visual layering and contrast of the image, making the content clearer and brighter, thereby improving the student's learning experience and the effectiveness of teaching content delivery.

As a common probability distribution, the mathematical properties of the Gaussian distribution make it well-suited for describing the color distribution in an image. For online classroom images, the histograms of the R, G, B color channels are often not perfectly uniform and may exhibit color biases or excessive concentration in certain areas. To adjust these uneven distributions, this paper uses a single Gaussian model to model the color distribution of each color channel. In this model, the expectation value ω of the Gaussian distribution is used to describe the central position of the color distribution, while the standard deviation δ controls the extent of the distribution's spread. Specifically, ω represents the average color position of each color channel in the image, i.e., the overall color tone center for red, green, and blue; δ determines the concentration of the color distribution. A smaller σ leads to a more concentrated color distribution, reducing the image's color deviation, while a larger δ results in a more dispersed color distribution, making the image's color performance more diverse. The Gaussian distribution can be expressed as A~V(ω,δ²). If the random variable A follows a Gaussian distribution, its probability density function is:

$d(a)=\frac{1}{\sqrt{2 \tau} \delta} \exp \left(-\frac{(a-\omega)^2}{2 \delta^2}\right)$     (5)

Through this modeling, this paper can independently analyze and optimize each color channel in the online classroom visuals. In practical application, the first step is to use the Maximum Likelihood Estimation (MLE) method to solve for the mean ω and standard deviation δ of each channel, and then adjust the image based on these parameters. MLE is a commonly used statistical method that estimates the optimal parameter values by maximizing the likelihood function of the observed data. In this method, the ω and δ obtained by MLE accurately reflect the distribution characteristics of each color channel in the online classroom image, thus providing a precise basis for subsequent histogram stretching adjustments. Assuming that the estimated value of ω is represented by $\widehat{\omega}$, the estimated value of δ is represented by $\hat{\delta}$, the sample values and each pixel value are represented by A₁, and the average pixel value by $\bar{A}$, with the number of samples represented by v, the specific solving formula is:

$\hat{\omega}=\bar{A}=\frac{1}{v} \sum_{u=1}^v A_u$     (6)

$\hat{\delta}^2=\frac{1}{v} \sum_{u=1}^v\left(A_u-\bar{A}\right)^2$     (7)

3.2 Histogram stretching

In the visual enhancement of online classrooms, the clarity of images and the natural presentation of colors are crucial for improving teaching effectiveness and student learning experience. For online classroom images, this paper proposes a histogram stretching method based on Gaussian model modeling, aiming to improve the visual performance of the image through precise adjustments and optimization of color distribution, making the image more vivid and layered, thereby enhancing the transmission effect of the teaching content.

Firstly, the first step of histogram stretching is to perform histogram analysis on the compensated online classroom image and normalize the pixel values of the image, mapping the pixel values from the original range of [0, 255] to the standardized range of [0, 1]. The purpose of normalization is to eliminate brightness differences that may exist between different image sources and devices, so that the color range of each image is unified, thus providing a consistent basis for subsequent image adjustments. Further, this paper uses a Gaussian model-based modeling method to analyze each color channel in detail, estimating the expected values ω_e, ω_h, ω_y and standard deviations δ_e, δ_h, δ_y for each channel. Then, based on the maximum and minimum values of each color channel, the color extension coefficient βz is calculated. The purpose of this coefficient is to adjust the contrast and brightness of each color channel, making the color range of the image richer and more balanced. Let the maximum and minimum pixel values of the three channels be represented by U_zMAX and U_zMIN, and the calculation formula is:

$\beta_z=\frac{1}{U_{z M A X}-U_{z M I N}}$     (8)

After completing the above steps, for the R, G, and B color channels of the online classroom image, this paper makes the final adjustment through histogram stretching. In this process, each channel of the histogram is independently extended, allowing each color channel of the image to better cover the color space, thereby enhancing the color depth and layering of the image. Let the stretched R, G, and B channel values be represented by U_ez', U_hz', and U_yz', and the compensated R, G, and B channel values be represented by U_ez, U_hz, and U_yz, respectively. The expected values and standard deviations of the three-channel histograms are represented by ω_e, ω_h, ω_y and δ_e, δ_h, δ_y, and the extension coefficient is represented by x_z. The specific process expression is:

$\left\{\begin{array}{l}U_{e z}^{\prime}=X-\beta_e \times \frac{\delta_h}{\delta_e} \times\left(\omega_e-U_{e z}\right), U_{e z} \leq \omega_e \\ U_{e z}^{\prime}=X+\beta_e \times \frac{\delta_h}{\delta_e} \times\left(U_{e z}-\omega_e\right), U_{e z}>\omega_e\end{array}\right.$     (9)

$\left\{\begin{array}{l}U_{h z}^{\prime}=X-\beta_h \times \frac{\delta_y}{\delta_h} \times\left(\omega_h-U_{h z}\right), U_{h z} \leq \omega_h \\ U_{h z}^{\prime}=X+\beta_h \times \frac{\delta_y}{\delta_h} \times\left(U_{h z}-\omega_h\right), U_{h z}>\omega_h\end{array}\right.$     (10)

$\left\{\begin{array}{l}U_{y z}^{\prime}=X-\beta_y \times \frac{\delta_h}{\delta_y} \times\left(\omega_y-U_{y z}\right), U_{y z} \leq \omega_y \\ U_{y z}^{\prime}=X+\beta_y \times \frac{\delta_h}{\delta_y} \times\left(U_{y z}-\omega_y\right), U_{y z}>\omega_y\end{array}\right.$     (11)

In the visual enhancement of online classroom images, improving the contrast and detail representation of the image is crucial, especially when viewed on different devices and in various environments. The contrast of the image often affects the student’s viewing experience and information retrieval effectiveness. To address the low contrast problem in online classroom images, this paper further introduces a dual enhancement strategy combining Contrast Limited Adaptive Histogram Equalization (CLAHE) and L-channel enhancement to ensure the clarity and contrast of the image's details and colors.

In online classroom scenarios, the readability of handouts, text, and charts is crucial, and these contents are often hard to distinguish in low-contrast environments. The CLAHE algorithm divides the image into several sub-blocks and applies adaptive histogram equalization on each sub-block for local enhancement, thereby effectively improving the contrast in local regions of the image. Enhancing the L-channel is another key step. The L-channel represents the luminance information of the image, which is the main factor affecting the perceived contrast and brightness levels of the image. In online classrooms, the visibility of the teacher’s facial expressions and the teaching content is directly related to the luminance distribution. By enhancing the L-channel, the brightness contrast of the image can be effectively improved, especially in the areas of the teacher’s face or charts, ensuring that students can clearly see the image even under different lighting conditions. Let the compensated channel value be represented by U_ez(X), the pre-compensated channel value be represented by U_e(X), the green channel value be represented by U_h(X), the average value of the green channel and compensated channel be represented by $\overline{U_h}$, $\overline{U_e}$, and the gain coefficient be represented by x. The formula is as follows:

$\left[\begin{array}{l}A \\ B \\ C\end{array}\right]=\left[\begin{array}{lll}0.412 & 0.358 & 0.180 \\ 0.213 & 0.715 & 0.072 \\ 0.019 & 0.119 & 0.950\end{array}\right]\left[\begin{array}{l}R \\ G \\ B\end{array}\right]$     (12)

The spatial reference values of the ABC space are represented by A_v, B_v, C_v, and the equation is:

$\left\{\begin{array}{l}M=116 d\left(B / B_v\right)-16 \\ x=500\left[d\left(A / A_v\right)-d\left(B / B_v\right)\right] \\ y=200\left[d\left(B / B_v\right)-d\left(C / C_v\right)\right]\end{array}\right.$      (13)

For online classroom images, enhancing the L-channel is especially helpful for improving the visibility of details such as the teacher’s facial expressions, blackboard writing, and charts, thereby enhancing the clarity and depth of the teaching content. By converting the RGB image to the Lab color space and applying histogram stretching to the L-channel, with a range set between 0.01% and 99.9%, and setting the luminance values outside the range to 100 and 0, the luminance dynamic range of the image can be effectively expanded. This stretching operation not only enhances the image contrast but also avoids the grayscale unevenness caused by color deviation. Let the modified luminance value be represented by M'_[u][k], the pre-modified luminance value be represented by M_[u][k], and the maximum and minimum luminance values within the stretching range be represented by U_MAX and U_MIN, the equation is as follows:

$M_{[u][k]}^{\prime}=\left(M_{[u][k]}-U_{M I N}\right) \cdot\left[100 /\left(U_{M A X}-U_{M I N}\right)\right]$     (14)

It should be noted that although both CLAHE and L-channel enhancement effectively improve the contrast and clarity of the image, the low luminance areas may become darker during the enhancement process, especially in noise reduction, where the overall details of the image may be lost due to noise suppression algorithms. This means that while the contrast of the image is improved, there may be issues of detail loss or unclear shadow areas, especially when the image contains a significant amount of noise. Therefore, in practical applications, in addition to using CLAHE and L-channel enhancement, further noise suppression and detail restoration strategies may be needed to ensure that the high contrast does not lead to unnecessary visual discomfort or detail loss.

3.3 Homomorphic filtering

In visual enhancement of online classroom images, ensuring image detail clarity and brightness balance is crucial to improving the student learning experience. In particular, under different lighting conditions, online classroom images may experience loss of detail in darker areas, which impacts the readability of teaching content. This is especially true when the viewer is in a relatively dark environment or when the display quality of the device is suboptimal. In such cases, the teacher's handwriting, diagrams, and screen-shared content may be hard to see clearly if the dark areas of the image lack detail. To address this issue, this paper proposes the use of homomorphic filtering to reduce surface reflections in the image, thereby minimizing detail loss.

Homomorphic filtering begins by applying a logarithmic transformation to the image, which converts the image's brightness information into an additive form. This allows the image's luminance component to be divided into low-frequency and high-frequency parts. In the frequency domain, Fourier transformation is applied to decompose the image, and then a suitable frequency-domain filter is used for processing, as shown in the following equation. This method suppresses the low-frequency illumination component, reducing image brightness inconsistencies caused by environmental lighting changes, while enhancing the high-frequency reflection component to improve the image's detail and contrast. Let the filter used to suppress low-frequency components and enhance high-frequency components be represented by the transfer function G(i,n). The expression is given by:

$G(i, n) \operatorname{DDS}(\operatorname{LN}(d(a, b))) =G(i, n) \operatorname{DDS}(\operatorname{LN}(u(a, b))) +G(i, n) \operatorname{DDS}(\operatorname{LN}(e(a, b)))$     (15)

where, e_g and e_m are the amplification factors for high-frequency and attenuation factors for low-frequency components, respectively; z is the sharpening coefficient; and i and n represent the horizontal and vertical distances from the filter point to the filter center. The cutoff frequency is denoted as F₀. The expression of G(i,n) can be written as:

$G(i, n)=\left(e_g-e_m\right) \cdot\left(1-\exp \left\{-z\left[i^2+n^2 / F_p^2\right]\right\}\right)+e_m$     (16)

Homomorphic filtering effectively enhances the overall brightness balance of the online classroom image and improves detail visibility in darker regions. In environments with poor lighting or insufficient contrast, the teacher's facial expressions, teaching materials, and text and charts on the screen often appear blurred due to uneven reflections of light. Homomorphic filtering can eliminate the impact of environmental lighting on the image, making the reflective components more prominent and thereby improving the visibility of details in the image.

This research mainly focuses on improving image quality in online classrooms and proposes an integrated solution based on adaptive color correction and visual enhancement technologies. First, through adaptive color correction, the issue of color distortion caused by lighting changes and differences in display devices is successfully addressed, making the image colors more natural and vivid, effectively improving the visual experience in online learning. Second, through visual enhancement technology, the proposed algorithm highlights and strengthens the key content of the classroom, increasing interaction and engagement, helping students capture the focus of their learning more clearly, and promoting enhanced learning efficiency. The experimental results show that the proposed method significantly outperforms traditional algorithms across multiple evaluation metrics, validating its superiority in color restoration, detail enhancement, and image quality.

This research provides an effective solution for improving online classroom image quality, especially in environments with common issues such as uneven lighting and differing display devices, demonstrating good adaptability and effectiveness. By combining color correction with visual enhancement technology, the proposed method can significantly improve students' visual experience, enhance classroom interaction, and engagement, making it highly significant for improving the quality of online education. However, there are certain limitations in this research. First, the experiments mainly focus on specific image processing scenarios, and future research can verify the algorithm's generalizability and robustness in a wider range of online classroom scenarios. Second, although the proposed algorithm has achieved good results in improving image quality, further exploration is needed regarding its applicability in real-time processing and in different network environments, particularly its efficiency and stability in large-scale online education applications.

Future research can further expand the algorithm proposed in this paper, especially in terms of optimization for different devices and environmental conditions, to improve its versatility and real-time performance. Specifically, studies can explore how to further enhance color correction and visual enhancement effects using deep learning techniques, particularly in complex lighting and variable scenarios. Additionally, by integrating virtual reality and augmented reality technologies, future research can explore more immersive online classroom experiences, further enhancing student learning outcomes and classroom interaction. Lastly, optimizing the algorithm to balance computational efficiency and image quality while reducing resource consumption will be another important direction for future research.