Integrating Metaheuristic Algorithms and Minimum Cross Entropy for Image Segmentation in Mist Conditions

Presently, there are several problems with the artificial vision of the autonomous cars, one of these is the detection of objects in adverse conditions, a typical example of this problem is rain where visibility is limited or non-existent, in the same way, autonomous driving under sand or snow storms or in mist conditions, just to mention a few. The core of this chapter deals with the problem in mist conditions by improving image segmentation. These types of images represent a challenge that requires the application of efficient and robust computer techniques that enhance; the segmentation, contrast and can be clearly interpreted. This work proposes the use of metaheuristic algorithms (MA) combined with threshold technique such as minimum cross entropy (MCE) to applied to images under mist conditions. The results of the work are analyzed under statistical tests and to guarantee proper segmentation several state-of-the-art metrics were used.