In this paper we present a method to extract text lines in handwritten documents. Indeed, line extraction is a first interesting step in document structure recognition. Our method is based on a notion of perceptive vision: at a certain distance, text lines of documents can be seen as line segments. Therefore, we propose to detect text line using a line segment extractor on low resolution images. We present our extractor based on the theory of Kalman filtering. Our method makes it possible to deal with difficulties met in ancient damaged documents: skew, curved lines, overlapping text lines. . . We present results on archive documents from the 18th and 19th century.