Track-Index-Guided Sustainable Off-Road Operations Using Visual Analytics, Image Intelligence and Optimal Delineation of Track Features

Abstract

Visual-analytics-guided systems are replacing human efforts today. In many applications, movement in off-road terrain is required. Considering the need to negotiate various soft ground and desertic conditions, the beaten tracks of leading vehicles considered to be safe and suitable for guiding are used in such operations. During night, often, these tracks pass through low-contrast conditions posing difficulty in their identification. The maximization of track contrast is therefore desired. Many contrast enhancement techniques exist but their effectiveness varies as per the surrounding. Other than conventional techniques, the role of texture too becomes important for enhancing the differentiable track contrast. Gray-level co-occurrence matrix (GLCM)-based statistic measures are used here to evaluate the track texture. These measures are seen to improve the contrast of vehicle tracks significantly. A track-index-based technique is proposed to sort various images as per their effectiveness in increasing the track contrast. Different forms of track indices are proposed and compared. The proposed track index is seen as effective in sorting 88.8% of contrast images correctly. The proposed technique of creating and sorting images based on the contrast level is seen as a useful tool for improved fidelity in many difficult situations for making the off-road operations sustainable.