Design and Implementation of Adaptive Psychological Perception System Based on Edge-Cloud Collaboration and Distributed SVM Ensemble for Handwriting Analysis

DOI: https://doi.org/10.62517/jbdc.202601221

Author(s)

Yike Duan1, Jinbao Li1, Yinhui Gao1, Guangyan Wang2,*

Affiliation(s)

1School of Computer Science, Central China Normal University, Wuhan, Hubei, China 2School of Information Engineering, Tianjin University of Commerce, Tianjin, China *Corresponding Author

Abstract

To address the limitations of traditional handwriting analysis methods that rely on desktop-based offline processing and struggle to integrate with daily digital note-taking applications, this paper designs and implements a handwriting-based psychological trait analysis system based on edge-cloud collaborative architecture and a distributed support vector machine (SVM) ensemble. The system adopts an "edge interaction offloading, cloud intelligent inference" computational paradigm. The front-end is built using the Flutter framework with a drag-and-drop image analysis interface, while the back-end constructs a dual-track parallel feature extraction pipeline based on OpenCV for spatial-geometry and physical pen-pressure characteristics. For small-sample multi-label classification tasks, binary relevance method is introduced to decouple 8-dimensional psychological indicators, and a distributed perception ensemble consisting of 8 RBF-kernel SVM classifiers is deployed. Experimental results on a dataset of 1533 samples demonstrate that the system achieves an average prediction accuracy of 98.5%, representing a 24.1 percentage point improvement over lightweight CNN models. Ablation experiments verify that the distributed SVM ensemble can effectively learn the mapping relationship between handwriting features and psychological indicators.

Keywords

Handwriting Analysis; Edge-Cloud Collaboration; Support Vector Machine; Image Preprocessing; Multi-Label Classification

References

[1] Fakhrurroja H, Machbub C, Prihatmanto A S, et al. Multimodal Fusion Algorithm and Reinforcement Learning-Based Dialog System in Human-Machine Interaction. International Journal on Electrical Engineering & Informatics, 2020, 12(4). [2] Li C, Zhang K, Lin Q, et al. Major depressive disorder recognition based on electronic handwriting recorded in psychological tasks. BMC medicine, 2025, 23(1): 282. [3] Namboodiri A M, Jain A K. Online handwritten script recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2004, 26(1): 124-130. [4] Golubitsky O, Watt S M. Online stroke modeling for handwriting recognition//Proceedings of the 2008 conference of the center for advanced studies on collaborative research: meeting of minds. 2008: 72-80. [5] Gagiu D, Sendrescu D. Detection of personality traits using handwriting and deep learning. Applied Sciences, 2025, 15(4): 2154. [6] Blending I A. Crayon: Customized On-Device LLM via Instant Adapter Blending and Edge-Server Hybrid Inference. [7] Chen W, Chen S, Leng J, et al. A review of cloud-edge SLAM: Toward asynchronous collaboration and implicit representation transmission. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(11): 15437-15453. [8] Panwar S, Nain N. Handwritten text documents binarization and skew normalization approaches//2012 4th international conference on intelligent human computer interaction (IHCI). IEEE, 2012: 1-6. [9] Zhang L, Wang K, Wan Y. An efficient transformer–CNN network for document image binarization. Electronics, 2024, 13(12): 2243. [10]Kacker R, Maringanti H B. Personality analysis through handwriting. GSTF Journal on Computing (JoC), 2014, 2(1). [11]Pathak A R, Raut A, Pawar S, et al. Personality analysis through handwriting recognition. Journal of Discrete Mathematical Sciences and Cryptography, 2020, 23(1): 19-33. [12]Tian Y, Shi Y, Liu X. Recent advances on support vector machines research. Technological and economic development of Economy, 2012, 18(1): 5-33. [13]Chandra M A, Bedi S S. Survey on SVM and their application in image classification. International Journal of Information Technology, 2021, 13(5): 1-11.