Leak Detection of Natural Gas with Base on the Components of Color Spaces RGB and HSI Using Novelty Filter

This paper proposes a method of detecting natural gas leak in oil installations, applying pattern recognition techniques to digital images using the concept of novelty classifier with digital images originated from Closed-Circuit Television of installation. For this, it was developed six different types of novelty classifiers. The difference between these classifiers is the color component of color space (RGB or HIS) used for novelty classifier training. The results obtained are promising, with an accuracy of 98.52%.