Low-cost thermal scanner image enhancement by merging thermal and visual data

This paper demonstrates the application of different image processing techniques to process high resolution visual images and merge it with low resolution thermograms to improve its level of detail. The same idea is applied in commercially available thermal cameras (e.g. FLIR with MSX® technology). Low resolution thermograms considered in this paper were obtained from a thermal scanner with point infrared detector (Fig. 1) sensitive to long wavelength infrared spectral range. The proposed algorithms are Laplacian, Sobel operator, embossing and Gaussian differential blur (described in section 3). The authors processed 6 different thermograms to qualitatively assess obtained results. It was done in a statistical manner through a survey and revealed that both Sobel operator and embossing provided the most clear, detailed and unambiguous results (Fig. 5). Such algorithms may be applied for processing more channels in a multispectral, cost-effective system.