Abstract:In this article, we propose a memristor-based selective convolutional (MSC) circuit for salt-and-pepper (SAP) noise removal. We implement its algorithm using memristors in analog circuits. In experiments, we build the MSC model and benchmark it against a ternary selective convolutional (TSC) model. Results show that the MSC model effectively restores images corrupted by SAP noise, achieving similar performance to the TSC model in both quantitative measures and visual quality at noise densities of up to 50%. Note that at high noise densities, the performance of the MSC model even surpasses the theoretical benchmark of its corresponding TSC model. In addition, we propose an enhanced MSC (MSCE) model based on MSC, which reduces power consumption by 57.6% compared with the MSC model while improving performance.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in a high - density salt - and - pepper noise (SAP) environment, how to effectively remove the noise in the image. Specifically, the paper proposes a memristor - based selective convolutional circuit (MSC) to deal with SAP noise, and further proposes an enhanced MSC model (MSCE) on this basis to improve performance and reduce power consumption. ### Problem Background 1. **Characteristics of Salt - and - Pepper Noise** - Salt - and - pepper noise is a common impulse noise, which is manifested as randomly appearing white (maximum value 255) and black (minimum value 0) pixels in the image. - Under high - density noise (for example, more than 50%), the effects of traditional denoising methods (such as median filtering, mean filtering, etc.) are not good. 2. **Limitations of Existing Methods** - Convolutional neural networks (CNN) perform well in dealing with Gaussian noise, but their performance in dealing with SAP noise is not satisfactory, especially under high - noise density. - Existing memristor - based SAP noise removal circuits are usually only effective under low - noise density (about 10%). ### Core Contributions of the Paper 1. **Memristor - based Selective Convolutional Circuit (MSC)** - A memristor - based selective convolutional circuit (MSC) is proposed to remove high - density SAP noise. - The effectiveness of this circuit under different noise densities has been verified through simulation experiments, especially excellent performance under 50% noise density. 2. **Enhanced MSC Model (MSCE)** - An enhanced MSC model (MSCE) is proposed, with a 57.6% reduction in power consumption while maintaining performance improvement. - The MSCE model shows performance exceeding the theoretical benchmark under high - noise densities of 60% and above. ### Experimental Results - **Quantitative Evaluation** - Peak signal - to - noise ratio (PSNR) and structural similarity index (SSIM) are used as evaluation indicators. - Under different noise densities, the performance of MSC and MSCE models is better than that of the existing TSC model. - **Qualitative Evaluation** - The visual quality of the image is significantly improved. Especially under high - noise density, the MSCE model can better restore image details. ### Summary This paper solves the problem of high - density SAP noise removal by introducing a memristor - based selective convolutional circuit, and further improves performance and energy efficiency through an improved algorithm. This provides a new solution for image processing in high - noise environments.

Memristor-Based Selective Convolutional Circuit for High-Density Salt-and-Pepper Noise Removal

A very fast and efficient multistage selective convolution filter for removal of salt and pepper noise

High Density Noise Removal by Cascading Algorithms

Study of MRI Image Denosing Method Based on MCA and Context Model

Memristive system design for variable pixel G-neighbor denoising filter

MSEMG: Surface Electromyography Denoising with a Mamba-based Efficient Network

HSMF: hardware-efficient single-stage feedback mean filter for high-density salt-and-pepper noise removal

Pruning and quantization algorithm with applications in memristor-based convolutional neural network

Memristor-based in-situ convolutional strategy for accurate braille recognition

Memristor–CMOS Hybrid Circuits Implementing Event-Driven Neural Networks for Dynamic Vision Sensor Camera

A deep convolutional neural network for salt-and-pepper noise removal using selective convolutional blocks

An efficient full-size convolutional computing method based on memristor crossbar

Memristor-based multi-synaptic spiking neuron circuit for spiking neural network

NAMF: A Non-local Adaptive Mean Filter for Salt-and-Pepper Noise Removal

NAMF: A Nonlocal Adaptive Mean Filter for Removal of Salt-and-Pepper Noise

Adaptive Noise Reduction Scheme for Salt and Pepper

Enhancing Robustness of Memristor Crossbar‐Based Spiking Neural Networks against Nonidealities: A Hybrid Approach for Neuromorphic Computing in Noisy Environments

A convolutional neural networks denoising approach for salt and pepper noise

A Memristive Spiking Neural Network Circuit with Selective Supervised Attention Algorithm

Multiparent Fractal Image Coding-Based Methods For Salt-And-Pepper Noise Removal

Spontaneous sparse learning for PCM-based memristor neural networks