In this paper, we propose a novel method to address the nighttime single image dehazing problem. Estimation of the ambient illumination map and transmission map are the key steps of modern dehazing approaches. For hazy scenes at night, ambient illumination is usually not globally isotropic as a nighttime scene typically contains multiple light sources. Frequently, Light source regions and non-light source regions exhibit distinct color features. However, existing nighttime dehazing methods have been attempting to process these two regions based on identical prior assumptions. Moreover, the commonlyused local maximum pixel method tends to over-estimate the ambient illumination. These two drawbacks result in color distortions and halo artifacts around the light source regions in the output images. In this work, we present a pixel-wise alpha blending method for estimating the transmission map, where the transmissions estimated from dark channel prior (non-light source region) and the proposed bright channel prior (light source region) are effectively blended into one transmission map guided by a brightness-aware weights map. Based on the Retinex theory, a channel difference guided filtering method is proposed to estimate the ambient illumination, which produces a spatially variant low-frequency passband that selectively retains the high-frequency edge details. Extensive experiments on the benchmarks demonstrate that our method outperforms the state-of-the-art methods for nighttime image dehazing, especially in terms of color consistency and halo artifacts reduction in the dehazed images.INDEX TERMS Nighttime image dehazing, image restoration, alpha blending.

Epilepsy is a common and genetically heterogeneous disorder among children. Advances in next-generation sequencing have revealed that numerous epilepsy genes, helped us improve the understanding of mechanisms underlying epileptogenesis, and guided the development of treatments. We identified 39 candidate variants in 21 genes, including 37 that were pathogenic or likely pathogenic variants according to the American College of Medical Genetics and Genomics scoring system and two variants of uncertain significance that were considered causative after they were associated with clinical characteristics. Thirty were de novo variants (76.9%), and 20 variants had not previously been reported (51.3%). We obtained a diagnosis in 39 of the 141 probands (27.7%). The most frequently mutated gene was SCN1A; KCNQ2, KCNT1, PCDH19, STXBP1, SCN2A, TSC2, and PRRT2 were mutated in more than one individual; ANKRD11, CDKL5, DCX, DEPDC5, GABRB3, GRIN2A, IQSEC2, KCNA2, KCNB1, KCNJ6, TSC1, SCN9A, and SCN1B were mutated in a single individual. In addition, we detected a nonsense variant in a candidate gene KCND1 and considered it as a new candidate epilepsy gene, which needed further functional study. Consequently, large number of unreported variants were detected, diverse phenotypes were associated with known epilepsy genes. Changes in clinical management beyond genetic counseling were suggested.

Early Infantile Epileptic Encephalopathy (EIEE) presents shortly after birth with frequent, severe seizures and progressive disturbance of cerebral function. This study was to investigate a cohort of Chinese children with unexplained EIEE, infants with previous genetic diagnoses, causative brain malformations, or inborn errors of metabolism were excluded. We used targeted next-generation sequencing to identify potential pathogenic variants of 308 genes in 68 Han Chinese patients with unexplained EIEE. A filter process was performed to prioritize rare variants of potential functional significance. In all cases where parental testing was accessible, Sanger sequencing confirmed the variants and determined the parental origin. In 15% of patients (n = 10/68), we identified nine de novo pathogenic variants, and one assumed de novo pathogenic variant in the following genes: CDKL5 (n = 2), STXBP1 (n = 2), SCN1A (n = 3), KCNQ2 (n = 2), SCN8A (n = 1), four of the variants are novel variants. In 4% patients (n = 3/68), we identified three likely pathogenic variants; two assumed de novo and one X-linked in the following genes: SCN1A (n = 2) and ARX (n = 1), two of these variants are novel. Variants were assumed de novo when parental testing was not available. Our findings were first reported in Han Chinese patients with unexplained EIEE, enriching the EIEE mutation spectrum bank.

Low cost, high efficiency, price transparency, and timely settlement of transactions are required for direct transactions between electricity providers and consumers in the microgrids. So the blockchain technology and the continuous double auction mechanism for direct electricity trading have always been a hot topic in the field of microgrids.In order to further reduce the transaction cost of blockchain and increase the transaction efficiency, and to solve the problem of lack of privacy protection for continuous double auction in the existing scheme, a privacy protection scheme of microgrids direct electricity transaction based on consortium blockchain and the continuous double auction is proposed. In it, the combination of consortium blockchain technology and continuous double auction mechanism is applied to reduce costs and improve the efficiency of transactions. In the meanwhile, pseudonyms and pseudonym certificates are generated by fair blind signature technology to realize identity privacy in the continuous double auction. And decentralization and user identity traceability are achieved by using (t, n) threshold secret sharing technology which distributes and recovers the private key of a trusted third party. The theoretical security analysis shows that the privacy protection scheme has higher security. The simulation experiment shows that the consortium blockchain technology has lower cost and higher efficiency in this scheme. INDEX TERMS Microgrid electricity trading, consortium blockchain, fair blind signature, (t, n) threshold secret sharing, privacy protection.

The nonlinearity of light emitting diodes (LED) has restricted the bit error rate (BER) performance of visible light communications (VLC). In this paper, we propose model-driven deep learning (DL) approach using an autoencoder (AE) network to mitigate the LED nonlinearity for orthogonal frequency division multiplexing (OFDM)-based VLC systems. Different from the conventional fully data-driven AE, the communication domain knowledge is well incorporated in the proposed scheme for the design of network architecture and training cost function. First, a deep neural network (DNN) combined with discrete Fourier transform spreading (DFT-S) is adopted at the transmitter to map the binary data into complex I-Q symbols for each OFDM subcarrier. Then, at the receiver, we divide the symbol demapping module into two subnets in terms of nonlinearity compensation and signal detection, where each subnet is comprised of a DNN. Finally, both the autocorrelation of the learned mapping symbols and the mean square error of demapping symbols are taken into account simultaneously by the cost function for network training. With this approach, the LED nonlinearity and the interference introduced by the multipath channel can be effectively mitigated. The simulation results show that the proposed scheme exhibits better BER performance than some existing methods and further accelerates the training speed, which demonstrates the prospective and validity of DL in the VLC system. INDEX TERMS Deep learning, visible light communication, LED nonlinearity, orthogonal frequency division multiplexing.