Fuel Nozzle Spray Pattern Classifier

In this study an industrial problem of classification of faulty fuel nozzles is considered and a solution is proposed by analyzing their spray pattern through vision based algorithms. The proposed solution is more reliable, accurate, cheap, and descriptive as compared to the manual techniques which are time consuming and error prone. We capture the dependency of spray patterns on imaging parameters using direction dependent enhancement, adaptive filtering and statistical feature extraction. In this study directional features of spray patterns affected by various disorders are extracted and are then used for classification of different fuel nozzles using Euclidean distance classifier. Moreover nozzle spray patterns are processed for spray angle measurement.