A Multimodal Image Encoding-Driven Visual Analytics Model for Financial Risk Assessment

In the context of digital transformation, enterprise financial data increasingly exhibit multimodal characteristics, with image data containing crucial risk-related information for financial decision-making.However, traditional financial risk analysis methods often underutilize multimodal image data.Existing studies either focus on single-modality encoding strategies-overlooking the heterogeneity and complementarity of multimodal data-or solely extract global image features while neglecting critical local details.Moreover, there is a lack of systematic integration between directional and non-directional feature encoding strategies for financial imagery, limiting their applicability in complex financial scenarios.To address these challenges, this paper proposes a multimodal image encoding fusion framework tailored for financial risk visual analytics.A local encoding scheme is constructed based on the Weber Local Descriptor (WLD), integrating both directional and non-directional encoding strategies.This approach enables efficient encoding and feature fusion of multimodal financial image data.The experimental results demonstrate that the proposed model provides more precise visual representations for financial risk analysis, significantly enhancing risk identification accuracy and decision support capabilities.This work promotes the cross-disciplinary integration of image encoding techniques and financial risk management.