**5. Conclusions**

In this paper, our network can be divided into two parts: feature extraction and post-processing. The post-processing part using the progressive expansion algorithm can guarantee the accuracy of text detection, but the experimental results prove that the simple use of FPN network in the feature extraction part has insufficient feature extraction, which leads to the decline of the text detection effect. This paper proposes a new scene text detection method based on feature fusion. This method uses SENet as the basic network and integrates the features of the MPANet to make up for the lack of features extracted from the original network. The fusion strategy proposed in this paper enables the text detection model to reach a detection level higher than that of the original network. Finally, the progressive expansion algorithm is used for post-processing so that the entire model can detect the text quickly and accurately. With the aim of improving the experimental results, the method in this paper avoids the introduction of end-to-end networks with too many parameters, and finally achieves the purpose of accurate and fast text detection, which is of grea<sup>t</sup> significance for the research of natural scene text detection technology oriented toward actual application scenarios. Furthermore, I hope to introduce new mathematical tools for research and discussion. In that regard, a recent approach based on geometric algebra [48] extracts features for multispectral images to be investigated. Finally, other multi-dimensional data processing such as L1-norm minimization [49] and hashing networks [50] remain primarily unexplored and can benefit from further research.

**Author Contributions:** Conceptualization, S.L. and W.C.; methodology, S.L.; software, S.L.; validation, S.L.; formal analysis, S.L. and W.C.; investigation, S.L. and W.C.; resources, S.L. and W.C.; data curation, S.L. and W.C.; writing—original draft preparation, S.L. and W.C.; writing—review and editing, S.L. and W.C.; visualization, S.L. and W.C.; supervision, W.C.; project administration, W.C.; funding acquisition, W.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China under gran<sup>t</sup> 61771322 and gran<sup>t</sup> 61871186, and funded by the Fundamental Research Foundation of Shenzhen under Grant JCYJ20190808160815125.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.
