ABSTRACT

In this work, an oxide memristor was utilized for analog computing, which exhibited controllable transitions within its stable resistive states. The memristive ON and OFF states were considered to be logic 1 and logic 0, respectively, whereas the intermediate memristances represent the possible resistive states within the memristive state machine (MSM). This MSM was sprucely tailored to perform tunable edge detection. In contrast to other software based conventional edge detection systems such as Canny, Sobel, and Prewitt, the proposed MSM system performed better with around 8% improvement in accuracy of edge detection. With the level of simplicity and accuracy, the proposed system exhibits potential to be an efficient alternative to the conventional systems for edge detection and laid a concrete path for future low hardware based smart system design due to the suppression of analog-to-digital conversion.

摘要

在这项工作中，氧化物忆阻器被用于模拟计算，它在其稳定的电阻状态下表现出可控的转换。忆阻 ON 和 OFF 状态分别被认为是逻辑 1 和逻辑 0，而中间忆阻表示忆阻状态机 (MSM) 内可能的电阻状态。这个 MSM 是为执行可调边缘检测而精心定制的。与其他基于软件的传统边缘检测系统（如 Canny、Sobel 和 Prewitt）相比，所提出的 MSM 系统性能更好，边缘检测精度提高了约 8%。凭借简单和准确的水平，