A handwritten signature is a unique personal identifier. It plays an important role in validating legal, administrative, and official documents. As more processes move to digital formats, automatically detecting handwritten signatures in scanned multilingual documents is crucial for reliable authentication. However, this task is difficult. Traditional Optical Character Recognition (OCR) systems often have a hard time with cursive writing, overlapping strokes, and complex handwritten patterns. To solve this problem, a new two-stage technique is proposed. It starts with OCR-based text masking to remove printed content. Then, it uses Connected Component Analysis (CCA) to find handwritten areas that are likely to contain signatures. In the first stage, without any preprocessing, the system reached a detection accuracy of 81.71%. After introducing the preprocessing technique, which effectively isolates signature components, the accuracy rose to 87.86%. This method was tested on a custom dataset that included 500 handwritten document images in 10 different languages. It effectively deals with cases that have no signatures, one signature, or multiple signatures by choosing the most prominent one. This provides a practical and scalable solution for detecting signatures in noisy and varied handwritten document settings.