Generally, underwater optical imaging is seriously impeded by challenging underwater optical environments, mainly in two aspects: first, the wavelength-selective absorption and scattering effects that significantly reduce and distort the contrast between the object and the background, and second, the inhomogeneous illumination effect that generates false and deceptive appearances, giving the impression of being underwater objects themselves. To tame these issues, researchers have developed various approaches, among which, object segmentation has absorbed many interests.
Object segmentation is the process of assigning a label to every pixel in an image such that pixels with the same label share certain characteristics, objects or background. This approach has emerged favorite owing to the fact that it offers condensed and informative details that can be used for many aspects of underwater research. Of interest in underwater research is the ability to discern object morphology details. As of now, levels sets have been integrated within the object segmentation technique in a bid to improve it. However, such endeavors do not obtain satisfying results in underwater environments; one notable example being that it is difficult to ensure a proper initial step for underwater images. Moreover, underwater nature interfere the model convergence, because textural background likely makes deviation of the model evolution. These problems are especially outstanding in underwater environments.
In a recent publication, Hohai University scientists in China namely: Professor Zhe Chen, Dr. Nan Qiu, Professor Lizhong Xu and Dr. Yunbo Xiong in collaboration with Professor Hong Song at Zhejiang University introduced optical models to underwater object segmentation and developed a new novel optically guided level set method. Their focus was to mathematically formulate the optical principles in underwater active imaging, and combine them with the image segmentation model. This way, they hoped to transform optical challenges in underwater environments into valuable guidance for underwater object segmentation. Their work is currently published in the research journal, Optics Express.
Briefly, they commenced by reviewing existing literature so as to gather information regarding the problem statement and set a strong foundation for their work. All in all, they combined their novel optical collimation guidance with a level set model, and obtained a novel optically guided level set.
The authors reported that under a non-supervision pattern, the level set guidance automatically ensured that the initial contour correctly enclosed the objects and evolved toward the desired edges. Moreover, in contrast to existing image segmentation methods as well as the methods for salient object detection, their approach showed better performance for identifying the object region and object contour from underwater images.
In summary, a novel optically guided level set has been firstly presented by Professor Zhe Chen and colleagues for underwater object segmentation. The results demonstrated that the optical principles in challenging underwater environments provided important cues for overcoming the difficulties inherent in underwater conditions. Overall, their novel underwater idea for optical imaging and image processing would set precedence in areas as diverse as computer vision and act as a motivation for future research.
Zhe Chen, Nan Qiu, Hong Song, Lizhong Xu, Yunbo Xiong. Optically guided level set for underwater object segmentation. Volume 27, Number 6 |2019 | Optics Express 8819.Go To Optics Express