A common problem in signal processing is the detection of signals in noisy environments. The author presents a fractal model to detect signals where detection is linked to the geometric complexity of the signal. He illustrates the utility of this approach to different signal types such as sinusoids and speech. It is shown that this approach is extendible to detection of two-dimensional data, such as images, sonar, and radar.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>