Dehazing of color image using stochastic enhancement

Bhattacharya, Sumanta; Gupta, Sumana; Venkatesh, K. S.

doi:10.1109/icip.2016.7532759

Cited by 14 publications

(17 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fig. 2 shows probability densities of mean colour uniformity in thick haze and thin haze regions evaluated on database from [8] and standard test images taken from [4]. It can be observed from the figure that probability density of mean colour uniformity metric is highly centred for thick haze regions, while for thin haze regions, it has a wider spread.…”

Section: Colour Uniformity Principlementioning

confidence: 96%

“…Fattal [4] used colour line prior for natural images to estimate the transmission map and the dark channel prior to estimate the A. Berman and Avidan [6] introduced the concept of the haze line to calculate the transmission map and used Sulami's method [24] based on dark channel prior to estimate parameter A. Bhattacharya et al [8] estimated reflectance map from edge strength and formulated dehazing as an optimisation process to increase the contrast using stochastic technique.…”

Section: Previous Workmentioning

confidence: 99%

“…Methods based on dark channel prior [2,[18][19][20] are susceptible to errors due to high reflecting surfaces, and those based on edges, colour priors [4,6,8] are susceptible to errors in the presence of blur and also require large processing time. Thus, these methods are not suitable for realtime applications.…”

Section: Previous Workmentioning

confidence: 99%

“…Here, A i and Â i represent the ground truth and estimated values, respectively. The accuracy of estimated A is compared with state of the art Sulami's method [24], Berman's method [30], and Bahat's method [31] for database [8]. This database contains 56 natural images with thick haze, depicting a real-life scenario for a dehazing application.…”

Section: Rmsementioning

confidence: 99%

See 3 more Smart Citations

Realtime dehazing using colour uniformity principle

Kumar¹,

Gupta²,

Venkatesh³

2019

IET Image Processing

Self Cite

View full text Add to dashboard Cite

Section: Colour Uniformity Principlementioning

confidence: 96%

Section: Previous Workmentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

Section: Rmsementioning

confidence: 99%

See 2 more Smart Citations

Realtime dehazing using colour uniformity principle

Kumar¹,

Gupta²,

Venkatesh³

2019

IET Image Processing

Self Cite

View full text Add to dashboard Cite

“…Conventional methods use hand-crafted features to capture the statistical properties of hazy images for estimating the transmission maps and atmospheric lights. For example, Tan et al obtained a clear image by maximizing the per-patch contrast based on the prior that hazy images usually have lower contrast than that of clear images [ 9 ]. However, the halo artefacts and colour distortion usually appear in the dehazing results.…”

Section: Introductionmentioning

confidence: 99%

A Novel Residual Dense Pyramid Network for Image Dehazing

Yin

Wang

Yang

2019

Entropy

View full text Add to dashboard Cite

Recently, convolutional neural network (CNN) based on the encoder-decoder structurehave been successfully applied to image dehazing. However, these CNN based dehazing methodshave two limitations: First, these dehazing models are large in size with enormous parameters, whichnot only consumes much GPU memory, but also is hard to train from scratch. Second, these models,which ignore the structural information at different resolutions of intermediate layers, cannot captureinformative texture and edge information for dehazing by stacking more layers. In this paper, wepropose a light-weight end-to-end network named the residual dense pyramid network (RDPN)to address the above problems. To exploit the structural information at different resolutions ofintermediate layers fully, a new residual dense pyramid (RDP) is proposed as a building block.By introducing a dense information fusion layer and the residual learning module, the RDP canmaximize the information flow and extract local features. Furthermore, the RDP further learnsthe structural information from intermediate layers via a multiscale pyramid fusion mechanism.To reduce the number of network parameters and to ease the training process, we use one RDPin the encoder and two RDPs in the decoder, following a multilevel pyramid pooling layer forincorporating global context features before estimating the final result. The extensive experimentalresults on a synthetic dataset and real-world images demonstrate that the new RDPN achievesfavourable performance compared with some state-of-the-art methods, e.g., the recent denselyconnected pyramid dehazing network, the all-in-one dehazing network, the enhanced pix2pixdehazing network, pixel-based alpha blending, artificial multi-exposure image fusions and thegenetic programming estimator, in terms of accuracy, run time and number of parameters. To bespecific, RDPN outperforms all of the above methods in terms of PSNR by at least 4.25 dB. The runtime of the proposed method is 0.021 s, and the number of parameters is 1,534,799, only 6% of thatused by the densely connected pyramid dehazing network.

show abstract

Dehazing of color image using stochastic enhancement

Cited by 14 publications

References 23 publications

Realtime dehazing using colour uniformity principle

Realtime dehazing using colour uniformity principle

A Novel Residual Dense Pyramid Network for Image Dehazing

Contact Info

Product

Resources

About