Nonlinear computer image scene and target information extraction based on big data technology

Jiaqi WangNonlinear 2023 · DOI

Abstract

To explore the extraction of computer image scene and target information, a nonlinear method based on big data technology is proposed. The method can decompose the computer image into a plurality of components when the SAR computer image is processed such as target extraction and computer image compression, which represent different captured image features, respectively. Selecting the most suitable processing method according to the characteristics of different components can greatly improve the performance. Using nonlinear diffusion method, the computer image is decomposed into structural components representing large-scale structural information and texture components representing small-scale detailed information, and the automatic threshold estimation in the diffusion process is studied. The LAIDA criterion is introduced into the automatic threshold solution of nonlinear diffusion-based computer image decomposition to test and evaluate the diffusion process of various diffusion parameter forms. The results show that the experimental outcome of the diffusion decomposition based on automatic threshold estimation is very close on each index, which shows that using automatic threshold estimation, no matter what diffusion index is used, very close results can be obtained. Specifically, for each algorithm, the parameter estimation threshold l for outliers plays an obvious role. The third is the degree of initiative of the estimation process. The larger the L , the larger the outlier, which will lead to a greater extent of the diffusion process, resulting in a continuous decrease in the structural similarity index and compositional correlation. It is proved that the algorithm has strong global search ability, can effectively avoid premature convergence, has fast convergence speed, and good long stability. It can be widely used for optimization of various multimodal functions.

Introduction

In the era of "big data," the sheer volume of video and image information can be overwhelming. Whether for intelligent surveillance or satellite imagery analysis, the ability to automatically separate a target from its background is crucial. Traditional image processing often struggles with "scene clutter"—the mix of large-scale structures (like buildings or terrain) and fine-grained textures (like grass or noise). When processing Synthetic Aperture Radar (SAR) images or surveillance footage, manual thresholding for image smoothing is inefficient and often fails to adapt to different image regions. This research introduces a nonlinear diffusion method that automatically decomposes images into structural and texture components using a novel thresholding technique based on the LAIDA criterion.

Methodology

Nonlinear Diffusion Decomposition and LAIDA Criterion

The paper treats image decomposition as a statistical estimation problem. Nonlinear diffusion is used to split an image into two layers: a structural component (representing large-scale features) and a texture component (representing small-scale details). To make the diffusion process adaptive, the researchers introduced the LAIDA criterion. By using robust statistics—specifically the Median Absolute Difference (MAD)—the algorithm automatically identifies edges as "outliers." This prevents the diffusion process from over-smoothing critical boundaries while still effectively reducing noise.

Dual-Branch FCN Architecture

For salient target detection, the paper employs a network that processes both RGB and Near-Infrared (NIR) images in parallel branches with shared weights. This dual-input approach allows the model to leverage complementary information from different spectral bands, enhancing the robustness of target extraction in complex scenes.

The dual-branch FCN architecture for salient target detection, utilizing weight sharing between RGB and NIR branches.

PReLU Activation

To solve the "dead ReLU" problem, standard Rectified Linear Units (ReLU) were replaced with Parametric ReLUs (PReLU). Unlike ReLU, which zeros out all negative values, PReLU learns a small slope for negative inputs. This keeps neurons active, accelerating the learning process and improving the global search ability of the algorithm to avoid premature convergence.

Results and Discussion

The study utilized a self-built sample set for training, with images standardized to $320 \times 320$ pixels. The training regime consisted of 200,000 iterations on a GeForce GTX 1080 GPU.

Key findings include:

  • Stability Across Indices: Automatic threshold estimation proved remarkably stable. Regardless of the specific diffusion index used, experimental outcomes remained consistent, demonstrating the method's robustness.
  • Clutter Suppression: The performance was evaluated using the Residual Clutter Ratio (RCC). As shown in the analysis of the parameter $L$ (the threshold for outliers), increasing $L$ leads to a more extensive diffusion process, which effectively reduces the residual clutter in the structural component.
Impact of the parameter L on the Residual Clutter Ratio (RCC), demonstrating how the diffusion process affects clutter suppression.
  • Faster Convergence: The PReLU-based network reached lower loss values significantly faster than the standard ReLU version, a vital efficiency gain for big data applications.

Conclusion

The proposed nonlinear diffusion method, combined with a dual-branch FCN and PReLU activation, offers a robust framework for image scene analysis and target extraction. While the method excels at static target extraction, future work will focus on integrating temporal action labels and specialized action detection networks to move from static scene analysis to dynamic behavior understanding and improved image captioning.

Summary updated

Abstract

Abstract To explore the extraction of computer image scene and target information, a nonlinear method based on big data technology is proposed. The method can decompose the computer image into a plurality of components when the SAR computer image is processed such as target extraction and computer image compression, which represent different captured image features, respectively. Selecting the most suitable processing method according to the characteristics of different components can greatly improve the performance. Using nonlinear diffusion method, the computer image is decomposed into structural components representing large-scale structural information and texture components representing small-scale detailed information, and the automatic threshold estimation in the diffusion process is studied. The LAIDA criterion is introduced into the automatic threshold solution of nonlinear diffusion-based computer image decomposition to test and evaluate the diffusion process of various diffusion parameter forms. The results show that the experimental outcome of the diffusion decomposition based on automatic threshold estimation is very close on each index, which shows that using automatic threshold estimation, no matter what diffusion index is used, very close results can be obtained. Specifically, for each algorithm, the parameter estimation threshold l for outliers plays an obvious role. The third is the degree of initiative of the estimation process. The larger the L , the larger the outlier, which will lead to a greater extent of the diffusion process, resulting in a continuous decrease in the structural similarity index and compositional correlation. It is proved that the algorithm has strong global search ability, can effectively avoid premature convergence, has fast convergence speed, and good long stability. It can be widely used for optimization of various multimodal functions.
BibTeX
@inproceedings{wang2023nonlinear,
  author = {Jiaqi Wang},
  title = {{Nonlinear computer image scene and target information extraction based on big data technology}},
  booktitle = {Nonlinear},
  year = {2023},
  doi = {10.1515/nleng-2022-0245},
}
Figure