Application of Polarization Sensing for Detection in Scattering Media ^†

Abstract

In recent years, significant progress in polarization imaging technologies has been observed, resulting in a numerous applications of this technology in biomedical imaging, autonomous vehicle navigations, 3D surface inspection, and many others. One of the most important applications of polarizing imaging is improving the image quality in scattering media. A good example is a number of conducted research and development works on improving the quality of images in underwater vision, providing impressive application results. In this work, we focused, however, on an agriculture industry-oriented solution, addressing the challenge of high-speed, highly reliable detection of the pits in cherries. In particular, different setup configurations for polarization image analysis using liquid crystal (LC) filters were investigated, and the examination of the sensitivity of the polarization systems was performed. It should be noted here that the polarization imaging systems are usually less sensitive, and the acquired images are of insufficient quality. That is why machine learning technology was used to enhance the object detection efficiency, and the method of extracting the details of the acquired images and improving detection accuracy based on machine learning was presented.