Infrared Imaging Based on Compressive Sensing for Power Equipment Inspection

摘要

Infrared thermal imager has been broadly applied in power inspection. To increase the system quality, the target information should be obtained with as little as possible data. To obtain the target information by infrared imager with less data, a framework based on compressive sensing was studied here. By projecting the original target to a sparse space using wavelet transform, the target information can be represented sparsely. The measure matrix was constructed using Gaussian random distribution to position the sparse of the original target. An improved orthogonal matching pursuit algorithm was used to reconstruct the original target from the measurements. So a high resolution infrared image can be obtained by less linear measurements sampled by low resolution sensor. The results on several typical infrared targets of power inspection show that super-resolution infrared imaging can be realized by compressive sensing; the reconstructed images were precise enough compared with the original high resolution images. The mean square error between the reconstructed image and original image was low, and the PSNR is large.