Recently, a team led by Prof. QIU Yuehong from Xi'an Institute of Optics and Precision Mechanics (XIOPM) of the Chinese Academy of Sciences (CAS) proposed multi-brightness layers with a genetic optimization algorithm for stereo matching under dramatic illumination changes. Their up-to-date result was published on Applied Optics.
Stereo matching is a key point in stereo vision which can be used in many practical fields, such as self-driving car, and remote sensing. However, many existing methods were tested under similar light conditions whose performance is not robust when the illumination changes dramatically and unevenly.
To address the above problem, QIU and his team members proposed a novel multi-brightness layers with genetic optimization (MBLG) algorithm. The whole procedure was divided into two main parts: (a) mechanism of histogram multi-brightness layers, (b) genetic optimization algorithm.
According to the experiments results, the proposed method is superior to the state-of-the-art stereo matching methods in accuracy and stability. In the future, the proposed MBLG method will provide an inspiration for designing novel algorithm of stereo matching under dramatic illumination changes.
Architecture overview of the MBLG method. (Image by XIOPM)
Download：