Enhanced Algorithm of Automated Ground Truth Generation and Validation for Lane Detection System by $\text{M}^{2}\text{BMT}$

Automotive imaging is a recent trend in research to assist drivers and is finally moving forward to achieve the goal of designing a driverless car. Along with a state-of-the-art algorithm, a state-of-the-art validation framework is also a requirement to ensure the quality of the system. This paper proposes an enhancement of the ground truth determination for automated lane detection system. The approach of time slicing has been built up on the binary framework. However, the classical binarization algorithms are not found to be good enough to address the particular domain of lane detection in an unconstrained environment and varied scenarios of lane structures, including curvy and dashed lane marks. This paper proposes a novel binarization algorithm based on min-between-max thresholding (MBMT). The adaptive binarization addresses the issue of outlier rejection in an efficient way and handles the effect of shadow, illumination variation, and other factors in time-sliced images in an automated manner. Additionally, this paper identifies the limitation of the classical time-slice-based approach even with time MBMT for ground truth determination and addresses the same through the second level of adaptation by spatial MBMT. Finally, a complete mathematical model is presented to validate any arbitrary lane detection algorithm with respect to the ground truth determined through the said method of hybrid or modified MBMT or
$\text{M}^{2}\text{BMT}$
.