A method of SF 6 gas leakage detection for infrared imaging based on Infrared Patch-Image model

Abstract Due to the low contrast and low signal-to-noise ratio (SNR) of infrared imaging of SF6 (Sulfur Hexafluoride) gas, it is susceptible to environmental noise, resulting in low accuracy and a high false alarm rate for existing detection algorithms. We propose a single-frame detection method for infrared imaging SF6 gas leakage based on the IPI model. Due to the robust principal component analysis (RPCA) limitation in describing complex backgrounds, this method uses weighted nuclear norm minimization (WNNM) to better describe the background's low-rank characteristics and then solve the problem using the Accelerating Proximal Gradient (APG) algorithm. Finally, it is possible to accurately extract SF6 gas leakage points from the sparse target image using adaptive threshold segmentation. We conducted experiments on both indoor and outdoor SF6 gas leakage tracks. The results demonstrate that the algorithm can effectively reduce the false noise generated by the complex background edge under indoor experimental conditions with SF6 gas leakage of 0.04ml/min and a distance of 5m, while also improving leak detection accuracy.