A compensation textures dehazing method for water alike area
- PDF / 1,007,459 Bytes
- 16 Pages / 439.37 x 666.142 pts Page_size
- 44 Downloads / 201 Views
A compensation textures dehazing method for water alike area Jian Zhang1,3 · Feihu Feng1 · Wanjuan Song2,3
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract With the continual development of deep learning, the image processing in Internet of Things is the key technology. Nevertheless, many deep learning methods cannot deal with the special needs of Internet of Things, for example, the Internet of vehicles and ships for the traffic haze image. Particularly, haze removal in the water area, because of the influence of water vapor, is more difficult than that in the ordinary scene. And the dehazing of water area has practical value in shipping and aerial photography. Sensible dehazing effect can even ensure the safety of navigation. In this paper, a compensation textures dehazing method is presented for water alike scene. The motivation of this paper comes from the following observations. Dark channel haze removal method has a very real dehazing effect for ordinary scenes. However, due to the principle of the dark channel method, this dehazing method has a large deviation in the water alike area. Therefore, based on the classical dark channel method, this paper proposes three innovations. First, a dynamic priority method is designed. This method can calculate the priority order of patches according to the characteristics of the processed subject. Second, a compensation textures method is designed, which can compensate the special area according to the proposed priority method. Third, a new haze removal method is designed, which can effectively remove the haze of water area according to the proposed compensation textures method. The results of visual and quality experiment show that proposed method has a state-of-the-art dehazing result in the water alike area. Keywords Image dehazing · Water alike area · Compensation textures · Deep learning · Internet of Things
* Feihu Feng [email protected] Extended author information available on the last page of the article
13
Vol.:(0123456789)
J. Zhang et al.
1 Introduction In recent years, deep learning and the Internet of Things (IoT) are more and more widely used. Particularly in Internet of vehicles and ships, there is an urgent need for new technologies to promote the development of these fields. Although the deep learning method has been applied to these fields, it still has shortages when dealing with some special problems, for example, the impact of haze on autonomous vehicles and ships. Although many deep learning methods can remove haze effective, for special scenes, the result of these methods cannot meet the system requirements [1–4]. In the process of research on image haze removal method [5–9], we found a common problem. That is, the existing haze removal methods including deep learning haze removal methods have poor dehazing effect on the water area [10–12]. There is a huge departure between the processing result and the real scene in the water area. Through further study, we found the physical reasons leading to this. For a hazi
Data Loading...
