Detecting textual information in scanned documents and document images is an essential task for Optical Character Recognition (OCR) systems and layout analysis. This study introduces a hybrid method which merges the XY Cut algorithm with LayoutLM models (v2 and v3) to achieve accurate and context-sensitive detection and extraction of text from annotated document datasets. The XY Cut algorithm divides document images into logical regions through horizontal and vertical pixel projection analysis while LayoutLM extracts and classifies text content in each region based on textual spatial and visual features. The approach enhances the analysis of document structure especially when handling complex layouts which includes forms and tables. Testing the proposed method on the FUNSD dataset demonstrates how it surpasses traditional techniques in both accuracy and efficiency. The evaluation results prove that the hybrid method can enhance OCR capabilities while also improving document understanding and information retrieval for digital archiving and administrative automation tasks and unstructured document layout information extraction. Document analysis technologies have achieved a major breakthrough with this development.