Memristor-Based Image Enhancement: High Efficiency and Robustness
Ruohua Zhu, Zhi‐Ri Tang, Shizhuo Ye, Qijun Huang, Lijun Guo, Sheng Chang
Abstract
Due to many outstanding physical characteristics, memristors have attracted much attention from all over the world. As a tendency, memristor-based systems are beginning to be applied in various fields of image processing, such as pattern recognition and edge detection. For the first time, memristors are introduced to image enhancement in this work, which dexterously processes the images twice via memristors’ intrinsic properties. Adopting a coarse transmission map and nonlinear memristive characteristics, the algorithm is highly efficient, which enormously reduces the computational cost, and image quality assessment demonstrates that it maintains comparable performance with classical algorithms. Furthermore, the temperature effect and the memristor instability, namely, the device-to-device variations and the cycle-to-cycle variations, are taken into consideration, and the average of several stacked images is proven effective in relieving the influence of variations. We believe that this work can explore a new application for the memristor.