|
The dual-band spectral differential (DB-SD) infrared gas imaging system adopts two wideband longwave-pass filters, the non-overlapping spectral region of the two filters are located at the strong absorption peak of the gas, and the differential images are obtained by subtracting between those two filtered images. Wideband longwave-pass filters can ensure high irradiance. While the differential processing is equivalent to a narrow-band filter, which can enhance the contrast between gas and background, thus effectively improving the gas detection capability of the thermal imager. However, differential processing inevitably accumulates random noise, resulting in the deterioration of image quality. In addition, for differential images with small gas leakage, there are still some problems, such as weak contrast and unclear boundary. Therefore, this paper aims at denoising and enhancement differential images. Considering the temporal redundancy and gas motion characteristics of differential image sequence, a 3D spatiotemporal filtering denoising algorithm based on approximately K-nearest neighbor patch matching and optical flow is proposed to improve the image quality. In addition, gas contrast enhancement algorithm based on variance statistical characteristics is proposed to highlight the gas region, which is equivalent to improving the gas detection capability of DB-SD imaging systems.
|
