Experimental fuzzy diagnosis algorithm based on image processing for rail profile measurement

Rail identification and measurement is very important for safety, comfort and maintenance. Therefore, rail profile diagnosis system can be used to determine faults in railways. This paper presents and validates a fuzzy diagnosis method based on image processing. The proposed approach builds an evaluation and integration between images obtained from camera and depth data measured with laser meter. Features extracted with filtering and preprocessing algorithms from images and depth data are evaluated and integrated with a fuzzy system. Main contributions in this study are using of intelligent classifier based on fuzzy logic, integrating of angle information with depth data, and extracting of efficient features from image data. Experimental results are presented for a cross-section of one hand of a rail. However, the method is high accuracy, robust, and suitable for real time measurement in all railways. The proposed approach is validated with healthy and faulty data.