An efficient single image haze removal algorithm for computer vision applications
- PDF / 3,954,494 Bytes
- 25 Pages / 439.37 x 666.142 pts Page_size
- 118 Downloads / 176 Views
An efficient single image haze removal algorithm for computer vision applications Gaurav Saxena 1
& Sarita Singh Bhadauria
2
Received: 26 June 2019 / Revised: 8 June 2020 / Accepted: 21 July 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Atmospheric conditions induced by suspended particles such as fog, smog, rain, haze etc., severely affect the scene appearance and computer vision applications. In general, existing defogging algorithms use various constraints for fog removal. The efficiency of these algorithms depends on the accurate estimation of the depth models and the perfection of these models solely relies on pre-calculated coefficients through the training data. However, the depth model developed on the basis of these pre-calculated coefficients for dehazing may provide better accuracy for some kind of images but not equally well for every type of images. Therefore, training data-independent based depth model is required for a perfect haze removal algorithm. In this paper, an effective haze removal algorithm is reported for removing fog or haze from a single image. The proposed algorithm utilizes the atmospheric scattering model in fog removal. Apart from this, linearity in the depth model is achieved by the ratio of difference and sum of the intensity and saturation values of the input image. Besides, the proposed method also take care the well-known problems of edge preservation, white region handling and colour fidelity. Experimental results show that the proposed model is more efficient in comparison to the existing haze removal algorithms in terms of qualitative and quantitative analysis. Keywords Atmospheric light . Defogging . Image quality . Edge preservation . Contrast gain . Image visibility
* Gaurav Saxena [email protected] Sarita Singh Bhadauria [email protected]
1
Jaypee University of Engineering and Technology, Raghogarh, Guna, Madhya Pradesh 473226, India
2
Madhav Institute of Technology and Science, Gole ka Mandir, Gwalior, Madhya Pradesh 474005, India
Multimedia Tools and Applications
1 Introduction In the bad weather conditions sometimes it is required to take the images for surveillance [5], remote sensing [8, 9], traffic control [11], aerial imaging [37], image/video retrieval [7, 32], deep-sea imaging [28] and other purposes. These images may be degraded by the different density particles (e.g. haze or fog, smoke, rain streak, etc.) present in the bad atmospheric conditions. Generally, images captured by the camera are used to acquire the information for a specific purpose but image captured by the camera may be corrupted due to scattering medium generated by the different density particles. This cause, the purpose of the capturing the image may not be fulfilled. Therefore, an efficient single haze removal technique is required to dehaze the captured images for computer vision applications. Moreover, the less information is visible in degraded images due to availability of haze particles; the recovery of an original image wi
Data Loading...
