Video trajectory analysis using unsupervised clustering and multi-criteria ranking

Sekh, Arif Ahmed; Dogra, Debi Prosad; Kar, Samarjit; Roy, Partha Pratim

doi:10.1007/s00500-020-04967-9

“…The first one is to explore the clear image without reference for training the dehazing network [45,46,51,52]. The second is to explore video dehazing [47][48][49][50].…”

Section: Discussionmentioning

confidence: 99%

GGADN: Guided Generative Adversarial Dehazing Network

Zhang

¹

,

Song

²

2021

Preprint

0

View full text Add to dashboard Cite

Image dehazing has always been a challenging topic in image processing. The development of deep learning methods, especially the Generative Adversarial Networks(GAN), provides a new way for image dehazing. In recent years, many deep learning methods based on GAN have been applied to image dehazing. However, GAN has two problems in image dehazing. Firstly, For haze image, haze not only reduces the quality of the image, but also blurs the details of the image. For Gan network, it is difficult for the generator to restore the details of the whole image while removing the haze. Secondly, GAN model is defined as a minimax problem, which weakens the loss function. It is difficult to distinguish whether GAN is making progress in the training process. Therefore, we propose a Guided Generative Adversarial Dehazing Network(GGADN). Different from other generation adversarial networks, GGADN adds a guided module on the generator. The guided module verifies the network of each layer of the generator. At the same time, the details of the map generated by each layer are strengthened. Network training is based on the pre-trained VGG feature model and L1-regularized gradient prior which is developed by new loss function parameters. From the dehazing results of synthetic images and real images, proposed method is better than the state-of-the-art dehazing methods.

show abstract

“…The first one is to explore the clear image without reference for training the dehazing network [45,46,51,52]. The second is to explore video dehazing [47][48][49][50].…”

Section: Discussionmentioning

confidence: 99%

GGADN: Guided generative adversarial dehazing network

Zhang

¹

,

Qian

²

,

Song

³

2021

Soft Comput

4

0

View full text Add to dashboard Cite

Image dehazing has always been a challenging topic in image processing. The development of deep learning methods, especially the Generative Adversarial Networks(GAN), provides a new way for image dehazing. In recent years, many deep learning methods based on GAN have been applied to image dehazing. However, GAN has two problems in image dehazing. Firstly, For haze image, haze not only reduces the quality of the image, but also blurs the details of the image. For Gan network, it is difficult for the generator to restore the details of the whole image while removing the haze. Secondly, GAN model is defined as a minimax problem, which weakens the loss function. It is difficult to distinguish whether GAN is making progress in the training process. Therefore, we propose a Guided Generative Adversarial Dehazing Network(GGADN). Different from other generation adversarial networks, GGADN adds a guided module on the generator. The guided module verifies the network of each layer of the generator. At the same time, the details of the map generated by each layer are strengthened. Network training is based on the pre-trained VGG feature model and L1-regularized gradient prior which is developed by new loss function parameters. From the dehazing results of synthetic images and real images, proposed method is better than the state-of-the-art dehazing methods.

show abstract

“…In recent years, a series of studies have been conducted regarding trajectory clustering in various aspects [16,21,30]. Some research tried to improve the performance of clustering through similarity indexes or matrix [33].…”

Section: Trajectory Clusteringmentioning

confidence: 99%

Spatial–Temporal Clustering and Optimization of Aircraft Descent and Approach Trajectories

Yang

¹

,

Tang

²

,

Chen

³

et al. 2021

Int. J. Aeronaut. Space Sci.

View full text Add to dashboard Cite

This study presents a procedure for the spatial-temporal clustering and optimization of aircraft descent and approach trajectories. First, the spatial-temporal similarity between two trajectories is defined. Clustering analysis are conducted to identify the prevailing trajectories. The clustering centers obtained based on spatial-temporal distance are compared with those obtained based on the traditional Euclidean distance. Second, a multi-objective optimization model is established to minimize fuel consumption, aircraft emission and noise impact considering flight constraints. The Pareto solution that has the highest similarity with the prevailing trajectories is selected as the final optimized trajectory. The performance indicators of the optimized trajectory are compared with the average values of historic trajectories. It is found that travel time, fuel consumption and noise impact for the optimized trajectory are reduced by 5.34%, 0.96% and 11.86%, respectively. The percentages are 0.96%, 1.32%, 9.18%, 3.54% and 4.00% for CO 2 , SO x , NO x , CO and HC, respectively. Also, the performance indicators for the two clustering centers based on spatial-temporal distance are generally closer to average performance of original trajectories, as well as those of the optimized trajectories, as compared with the two clustering centers based on Euclidean distance. The spatial-temporal clustering methods may help to discover the valuable information that lies in those indicators associated with features reflected in time dimension.

show abstract

Video trajectory analysis using unsupervised clustering and multi-criteria ranking

Cited by 17 publications

References 48 publications

GGADN: Guided Generative Adversarial Dehazing Network

GGADN: Guided Generative Adversarial Dehazing Network

GGADN: Guided generative adversarial dehazing network

Spatial–Temporal Clustering and Optimization of Aircraft Descent and Approach Trajectories

Contact Info

Product

Resources

About