Scene text detection plays a crucial role in the development of computer vision. However, current scene text detection algorithms mainly focus on Chinese and English, while Vietnamese scene text detection still remains a challenging task. The current algorithms to detect Vietnamese scene text frequently result in the incapacity of detecting Vietnamese diacritics and wrongly detecting background as text. To address these challenges, in this paper, a Vietnamese scene text detection algorithm is proposed to concentrate on diacritics and effectively reduce background interference. Specifically, an Edge Information Enhancement Module (EIEM) is first proposed to enhance the edge features of Vietnamese characters by combining a gradient filter with an attention mechanism. Secondly, a Text Region Enhancement Module (TREM) is proposed to enhance the feature representation of text regions by capturing global contextual information and dependencies among Vietnamese characters, thereby enhancing the distinction between background and text. Experiments on the Vintext dataset illustrate that the proposed method performs better in Vietnamese scene text detection tasks compared with several contemporary scene text detection algorithms. The code of the proposed algorithm is available at https://github.com/mlmmwym/VSTD-EITRFE.