We calculate photometric redshifts from the Sloan Digital Sky Survey Data Release 2 (SDSS DR2) Galaxy Sample using artificial neural networks (ANNs). Different input sets based on various parameters (e.g. magnitude, color index, flux information) are explored. Mainly, broadband photometry utilized and their performances in redshift prediction compared. While any parameter may be easily incorporated input, our results indicate that dereddened magnitudes often produces more accurate than...

Aiming at the missed detection problem in Single Shot MultiBox Detector (SSD), this paper proposes Contour feature fusion SSD algorithm (C-SSD). The performs operation on features layer to complement object contour, improve expression ability of information. This can enhance capability and reduce rate. simulation shows our C-SSD has a better performance detecting road participants such as vehicles, bicycles pedestrians compared with SSD. mAP ours is increased by 0.78 points test result...

Abstract To solve the current security problems of microservice architecture, mechanism API gateway platform using is designed and implemented. corresponding problems, a new designed. In data transmission, hybrid encryption algorithm hybridized by Rivest Shamir Adleman (RSA) Advanced Encryption Standard (AES) used for encryption. An between client introduced as only entrance users to access microservice, avoiding complexity logic caused direct communication server. After introduction...

In this paper, to better solve the problem of low tracking accuracy caused by sudden change target scale, we design and propose an adaptive scale mutation algorithm using a deep learning network detect first then track it kernel correlation filtering method verify effectiveness model through experiments. The improvement point paper is traditional at same time combine with apply in process tracking; addition not only can learn more accurate feature representation but also effectively cope...

With the development of big data technology and increasing demand for massive storage analysis, there are more mature distributed systems. In era data, traditional databases centralized technologies have problems such as low processing efficiency slow speed when which cannot meet needs modern information processing. Big is a new type that enhances capacity database, improves accelerates meets in data. Based on analysis respective characteristics architecture HBase Hive, this paper proposes...

Most pedestrian detection algorithms only provide the object region instead of actual body segmentation in video. For reducing large number redundant information and extracting a clear contour texture feature an up-right person, superpixel algorithm with correlation saliency analysis is proposed from coarse to fine cutting without any prior information. This cuts single target automatically obtains strict features human regardless whether or not camera moving. It exploits find location...

In view of the e-commerce platform merchants need to upload clothing labels system, but manual labeling appeared cumbersome information error, error-prone and other problems, this paper proposes a ResNet50-DS classification algorithm automatically label by using machines, complete clothing. Taking residual network as basic structure, SE-Net module is embedded in model, dropout layer added reduce generalization error model. The experiment shows that improves accuracy classification, which can...

For the human-being's body contour distortion problem of pedestrian object segmentation at complex indoor and outdoor environment for mobile robot visual applications, a cascaded superpixel algorithm was proposed considering background interference. Based on acquiring global blocks with primary computation, secondary achieved correlation degree average color center point Euclidean distance each between inside outside saliency detection region, in order to obtain upright person. With...

In order to further improve the frequency bandwidth efficiency and power of existing Continuous Phase Modulation (CPM) demodulation systems, a continuous phase modulation method based on Prolate Spheroidal Wave Functions (PSWFs) signal is proposed, which called CPM Based PSWFs (CPM-PSWFs). this method, space has with high energy aggregation in time-frequency domain flexible controllable time-bandwidth product as baseband pulse generate signal. At receiving end, coherent detection waveform...

The research difficulties of speech detection in the present focus on cases that signal-to-noise ratio(SNR) is low and background noise changes dramatically. For problem under SNR environments, based sparsity frequency domain sparse representation ability over-complete Fourier basis, signal reconstructed with Matching Pursuit algorithm, we propose a method which uses short time energy as feature. experimental results show this algorithm exhibits higher robustness white environments.

Compared with other virtualization technologies, container technology is widely used in edge computing because of its low cost, high reliability, flexibility and fast portability. However, the use can alleviate pressure massive data, but also bring complex diverse security problems. Reliable information risk assessment method key to ensure smooth application technology. According theory, a for containers based on dynamic game theory proposed. Aiming at attack defense process, system's model...

To decrease the computational complexity of adaptive inter-layer prediction and improve encoding efficiency in spatial temporal scalable video coding, a mode decision algorithm is proposed by exploiting part used modes co-located reference macroblock(s) for Hierarchical-B pictures all layers. This scheme generates reduced dynamic candidate list current macroblock according to statistical information derived from macroblocks their present level. The experimental results show that this...

A fast mode decision algorithm is proposed to solve the computation complexity for enhancement hierarchical-B pictures with adaptive inter-layer prediction in H.264/AVC scalable extension, video coding. This method exploits evaluated modes of co-located reference macroblocks, which include best and other abandoned modes, compose a new candidate group current macroblock motion estimation according their RD cost values. The can be reduced by adjusting size order decrease calculation....

As a non-sinusoidal modulation, the time domain orthogonal modulation based on Prolate Spheroidal Wave Functions (PSWFs) has advantages of high spectral efficiency and power utilization rate. However, in actual wireless communication systems, there are always non-ideal interferences such as synchronization deviation, multipath propagation, Doppler frequency shift. These factors have serious impact performance systems PSWFs. In this paper, from two aspects offset symbol offset, effect error...