Stochastic method of image edge detection using magnetic tunnel junction.

• Utilizing telegraphic switching signal of MTJ for stochastic computing. • Roberts Cross operator designed by stochastic computing. • Roberts cross operator realized by simple logic using MTJ. • Edge detection results differ based on generated bit-stream length. A hardware method for image edge detection is proposed by using stochastic behavior of magnetic tunnel junction. Telegraphic switching property of magnetic tunnel junction is employed for obtaining random bits which allows constructing a circuitry of a probabilistic bit-stream generator for stochastic computing. By applying the Roberts cross operator, we develop a stochastic way for edge detection with approximation of image gradient, which successfully conducts the role of image edge detection. Different edge detection results for different bit-stream lengths are shown in order to suggest the optimal bit-stream length for operation. [ABSTRACT FROM AUTHOR]