Authentication of Thermal Tomographic Images using Histogram Similarity Detection Techniques

The thermal tomography is gaining importance now a days because it is a very convenient, versatile and non-contact method which has been used for assessment of heating quality in a bounded volumetric space (e.g. furnace) subjected to heating. Thermal images generated thus experimentally, are then analysed using image processing technique for performance evaluation of heating quality (e.g. Uniform, Non-uniform etc.). In the pre-processing stage, the colour (R-G-B) images are extracted to three channel images (i.e. R, G and B) containing spatial pixel intensities. The images corresponding to each channel are stored as pixels values, each pixel value represents a colour intensity value. Histograms are frequency distribution (counts) of these intensity values that occur spatially. During training phase, the thermal tomographic image histogram for each channel (R, G and B) of desired heating quality are generated and stored. In testing phase, the unknown thermal tomographic image histogram is compared with trained histograms following histogram similarity method, which computes the overlapping area between the trained histogram and unknown histogram curves. The result shows that the area reaches maximum values for similar thermal images. The likeness for the similarity of the test image as captured are found out histogram similarity factors i.e. area of overlapped histogram curves, histogram-distance and skewness-kurtosis based similarity between the histograms. Heating quality (e.g. uniform or non-uniform) is then assessed based upon these found out histogram similarity factor values viz. an unknown test image has got the most similarity with that known heating class database which has got the maximum Histogram intersection percentage, minimum mean Euclidean distance between the normalised histograms and minimum mean Distance (Skewness-Kurtosis) measure.