ABSTRACT The existing image dehazing algorithms perform suboptimal in non‐homogeneous and/or dense haze scenarios. loss of feature information and alteration color distribution cause images to deviate from real‐world scenes when suppresses details. To address these issues, we design a dual‐branch network integrating discrete wavelet transform (DWT), multi‐scale fusion, constraints achieve dehazed with more realistic edges, textures, colors. Specifically, first introduce DWT into...

We propose two novel methods to improve the source location privacy security protection and node energy utilization in Wireless sensor networks (WSN). A preservation protocol for WSN based on angles(APS) an Enhanced (EAPS), which dynamically adjusts emission radius during routing. The APS produces geographically dispersed phantom nodes utilizes from energy-abundant regions make routing path versatile among entire network. In EAPS protocol, according number of its own adjacent nodes, residual...

In the Qinba mountainous area of Central China, pig farming has a significant impact on growth rural economy and substantially increased farmer incomes. Traditional knowledge plays an important role in selection forage plant species for by local people. This study aimed to identify plants used feeding catalog indigenous regarding their use. During 2016 2017, ethnobotanical surveys inventories were conducted Zhuxi County, Hubei Province, China. Data collected using semi-structured interviews,...

In order to solve the problem that neural network learns well but predicts badly, genetic algorithm was adopted optimize network. The LM-BP very well, and it is sensitive initial weights thresholds. Then its thresholds were selected by algorithm. So method of improving generalization ability for based on proposed. By example analysis, compared with randomly, optimized has high fitting precision testing accuracy. new can greatly improve

Aimed at improving the security and efficiency of anonymous authentication in vehicular ad hoc network (VANET), a certificateless batch scheme without bilinear pairings is put forward. By coordinating Trust Authority (TA) vehicles to generate public/private key pairs pseudonyms, system freed from dependency on tamperproof devices. Through comprehensive analyses, this proved not only be able realize such properties as authentication, anonymity, traceability, unforgeability, forward or...

The Aquila optimization (AO) algorithm has the drawbacks of local and poor accuracy when confronted with complex problems. To remedy these drawbacks, this paper proposes an Enhanced aquila (EAO) algorithm. avoid elite individual from entering optima, opposition-based learning strategy is added. enhance ability balancing global exploration exploitation, a dynamic boundary introduced. elevate algorithm’s convergence rapidity precision, retention mechanism effectiveness EAO evaluated using...

In this paper, a visual saliency model is used to detect the traffic signs in natural images. A number of features consist intensity, color opponency and orientations are extracted from image by model. Without need for segmentation, map indicating area where can be computed. From experiments on real scene images with domestic signs, detection rate 89% achieved.

In order to solve the problem that membership functions of traditional fuzzy control are difficult be determined, a new method (fuzzy network) is proposed. Fuzzy network adopted structure in which functions, rules, reasoning and defuzzification process included. Compared with neural network, don’t need learning training, output can directly obtained after data inputted. Through contrast design for automobile anti-collision system, not easy determined by method. Furthermore, local surfaces...

Acquirement of real-time and overall traffic information is very important for improving road network efficiency reducing congestion. This paper proposed an improved non-parametric background model to segment the moving vehicles from videos with limited computational complexity space complexity. With analysis characteristics parameters, a two-level classifier automatic recognition information. The results have high coincidence rate those expert classification.

In the social internet of things, community structure exists objectively and affects transmission network messages. If context such as is fully utilized, efficiency data forwarding will be effectively improved. A community-based routing algorithm (MSAR) proposed by studying multiple relationships. First, we propose four measures They are closeness degree, in-community activeness, cross-community activeness interaction. Then, design considers two stages. One other forwarding. The measurement...

The development of degree programs is not achieved overnight, but also has a certain internal growth cycle and stage. Drawing inspiration from the life theory, four-stage lifecycle theory with incubation period, differentiation period was proposed. Then, combining eight key tasks related to program construction, including cultivation, application, special qualification evaluation, periodic discipline level evaluation professional degree, quality inspection, dynamic adjustment, this paper...

The article analyzes the existing system for evaluating construction of degree programs, sorts out requirements various work tasks such as application, assessment, evaluation, and inspection, extracts data information related to direction, teachers, students, teaching research, process, quality, other aspects program construction. On this basis, a platform scheme was designed using fusion thinking, taking into account all relevant construction, unifying management This has positive promoting...