Image-Based Monte-Carlo Localization With Information Allocation Logic to Mitigate Shadow Effect

The development of vision-based navigation algorithms using a camera is becoming more important. The vision-based navigation can be categorized into two types. The first is to use sequential camera images as relative navigation. The second is to estimate the absolute navigation solution using a camera image and database. In absolute navigation, the difference between the database and the camera image is a major obstacle to image registration. One of the factors that make a lot of difference is the shadow effect. This shadow increases the inconsistency between the two images and eventually degrades the localization accuracy. This means that shadows have a significant impact when measuring the similarity of the two templates. To mitigate this effect, we inherited and developed the Monte Carlo Localization (MCL) algorithm based on a new similarity cost function, which is a key contribution to this article. We have established the importance of information with information reallocation logic that considers shadow areas. The proposed algorithm allocates the importance of the information considering a portion of the shaded area in the camera image. First of all, we analyzed the effects of shadows on the camera. To compare the performance of the algorithm, we used not only the shadow restoration algorithm but also various template-based matching algorithms. The proposed algorithm is validated through various simulations and real flight experiments as well.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.