Noise reduction in spatial data using machine learning methods for road condition data

Abstract

With the increase in the road transportation system the safety concerns for the road travels are also increasing. In order to ensure the road safety, various government and non-government efforts are visible to maintain the road quality and transport network system. The maintenance of the road condition is in the verse of getting automated for the quick identification of potholes, cracks and patch works and repair. The automation process is taking place in majority of the counties with the help of ICT enabled frameworks and devices. The primary device used for the purpose is the geo location enabled image capture devices. Regardless to mention the image capture process is always prune to noises and must be removed for better further analysis. Also, the spatial data is collected from the road networks are also prune to various error such as missing values or outliers due to the induced noises in the capture devices. Hence, the demand of the current research is to purpose a complete solution for the noise identification and removal from the spatial road network data for making the automation process highly successful and highly accurate. In the recent time, many parallel research attempts are observed, which resulted into solving the problem of noise reduction in all aspects of spatial data. Nevertheless, all the parallel research outcomes have failed to provide a single solution for all the noise issues. Henceforth, this work proposes three novel algorithms to solve spatial image noise problem using the adaptive moment filtration, missing value noise from the spatial data using adaptive logistic analysis and finally, the outlier noise removal from the same spatial data using corrective logistic machine learning method. The outcome of this work is nearly 70% accuracy in image noise reduction, 90% accuracy for missing value and outlier removal. The work also justifies the information loss reduction by nearly 50%. The final outcome of the work is to ensure higher accuracy for road maintenance automation.