Redundancy Data Elimination Scheme Based on Stitching Technique in Image Senor Networks

Transcription

1 Sensors & Trnsducers 204 by IFSA Publishing, S. L. Redundncy Dt Elimintion Scheme Bsed on Stitching Technique in Imge Senor Networks hunling Tng hongqing Technology nd Business Institute, No., Hulong Avenue, Jiulong Science nd Technology Prk, Jiulongpo District, hongqing ity, , hin Tel.: , fx: E-m: Received: 2 rch 204 /Accepted: 30 Apr 204 /Published: 30 June 204 Abstrct: Using imge stitching technology to find out overlp res in the cmer node cpture imge, inform the cmer node, cmer node wl fter redundncy imge dt which is overlp with djcent nodes, nd then cluster nodes rebud imges which hs been received, stitch together, form pnorm. Through this process, it cn reduce imge redundncy dt ffectivity in densely deployed wireless cmer imge sensor network, the network life cycle cn be gretly improved. This pper proposes new clcultion method. opyright 204 IFSA Publishing, S. L. Keywords: Imge sensor networks, Suture technique, Redundnt dt.. Introduction With the development nd socil economic progress, especilly in science nd technology constntly updted, the development of microprocessor technology nd sensor technology is fster nd fster. In support of these dvnced technologies, we cn induce sensor networks effectively cpture vriety of multimedi informtion, such s udio, video, imges, etc., nd grdully produced wireless multimedi sensor network []. With the prctice nd study, we found tht wireless multimedi sensor networks node is often only very short life cycle, it is becuse wireless multimedi sensor networks need to collect lot of informtion, nd require lot of computtion, which would gretly shortens the survivl of node cycles. This problem hs roused people's ttention in wireless multimedi sensor networks, using processors nd sensors, nd wireless communiction module only low power consumption, the use of bttery with lrger cpcity. So in this cse, very urgent problem is to tke series of mesures to chieve bttery life nd improved network lifetime purpose. urrently, reserch in this re, there hs been lot of reserchers hve conducted some reserch nd nlysis, series of mesures to chieve energy efficiency whe incresing the lifetime of network nodes. In this pper, on the bsis of these studies, the innovtion nd development, we propose new lgorithm, the imge stitching technology pplied to the imge sensor networks, the overlpped portion of the imge to cler, so the mount of dt wl be effectively reduced, communictions nturlly lower energy consumption nd prolong network lifetime.

2 2. Reserch Bckground Wireless Sensor in the dt collection process, there re mny cses wl result in the genertion of redundnt dt, first redundnt dt exists in the sme multimedi sensor node collected t the sme time between successive imges. Second, it is djcent to the multimedi sensor node t the sme time to the sme perception interest re of monitoring, there re redundnt dt between the collected imges. Thirdly, it is redundnt dt exists in the multimedi sensor nodes collect between djcent pixels in n imge. To solve these problems, we need to tke series of prompting, to eliminte the redundnt dt, or if the ggregtion node to send multiple multimedi sensor nodes to collect ll the dt, wl be on serious wste of communiction bndwidth, nd becuse of the need trnsfer lrge mounts of dt, the entire network wl consume too much energy, the life cycle of the network hve n dverse impct on the scle of network pplictions dversely ffected. For this problem, the imge stitching techniques cn be used to resolve. Imge stitching technology is wireless multimedi sensor networks is very importnt to dt processing technique. Through prctice studies hve shown tht the ppliction of wireless multimedi sensor imge stitching technology tht cn hndle from node in the network to obtin vriety of rw dt, which wl be prt of the complex to the ctive redundncy eliminted, the network dt trffic hs been effectively reduced energy consumption of this node cn be effectively reduced, the life cycle of the network cn lso ply role in prolonged. 3. oncept of Dt Fusion Becuse there re reltively dense node deployment in wireless multimedi sensor networks, so if multimedi sensor hs nerby loction, the informtion collected my be repeted nd redundnt. If you wnt to trnsfer ech node ggregtion lyer dt collected, it wl wste lot of communiction bndwidth. Also, lrge mounts of dt, but lso the mount of energy consumed by the network, the network lifetime effects. For this problem, dt fusion techniques cn be used, it refers to the multimedi informtion within the network collection process, the use of multimedi sensor node's locl computing nd storge cpcity to hndle this informtion collected, tht repet redundnt informtion to eliminte, so tht you do not need those unnecessry dt trnsmission, the nodes cn be effectively trnsmitted energy svings. In ddition, comprtive nlysis of multimedi sensor nodes cn collect informtion tht wl individul nodes to which there is lrge collection of informtion to the exclusion of error, so tht you cn improve the relibity of monitoring informtion. 4. The Advntges of Dt Fusion Technology in the Appliction of Imge Sensor Network Through prctice reserch shows tht the imge sensor to eliminte redundnt dt network spects of using suture techniques hs gret dvntges, embodied in three spects: First, the network cn be lrge degree of energy svings: Through the bove nlysis, we hve lerned tht the nrrtive, becuse the wireless multimedi sensor networks is reltively dense deployment node, then if multimedi sensor node hs n djcent loction, collected dt my be repeted nd redundnt; if ll these collected dt to the sink node, the node wl be gret wste of energy, nd therefore need to del with the redundnt dt. The suture techniques for dt fusion technology is n importnt one, cn be effectively pplied to the imge sensor to eliminte redundnt dt mong the network, the bove -mentioned lgorithm cn be used to eliminte the imge sensor network nodes redundnt dt collected, such mount of dt needs to be trnsferred hs been gretly reduced, multimedi sensor nodes hs been effectively trnsmit energy svings, nd thus the entire network for effective energy svings nd prolong the network lifetime. Second, the ccurcy of dt collection cn be enhnced: Becuse usully deployed wireless multimedi sensor network environment is unttended, then the hrsh environment on the node itself my ffect the norml opertion of fetures, some exceptions, report n error dt. Thus, single multimedi sensor nodes collect dt on the possible existence of errors, nd the use of dt fusion technology cn be the sme object multiple multimedi nodes collect dt for monitoring nd nlysis of the results, the error dt is excluded, thereby incresing dt ccurcy nd relibity. Third, the dt collection efficiency cn be significntly improved: we hve lredy mentioned bove, if the multimedi sensor node hs nerby loction, then there wl be lot of redundncy exists in the dt collected on these unnecessry the dt to be trnsmitted, the communiction bndwidth wl result in serious wste, nd the network wl be limited dditionl energy consumption. Thus, the need to use the dt fusion technique to the mount of dt to be trnsmitted reduced, so tht the burden of network trffic cn be gretly reduced, network trffic efficiency cn be improved, the trnsmission dely hs been gretly reduced, to chieve dt collection the purpose of improving efficiency. 5. Overview of Imge Stitching Technology Imge stitching technology s new technology, referring to the stitching multiple imges, mking it 2

3 whole lot of pnorms, these imges re the existence of sptil correltion. Then it cn be in the wireless multimedi sensor networks ppliction of this technology, in order to set the cmer node trnsmits those to eliminte redundnt imge dt, serve to reduce trnsmission energy consumption purposes []. Literture [2] for the first time in wireless multimedi sensor networks introduces imge stitching technology, simultion results show tht the introduction of suture technique over the imge, the imge sensor hs reltively dense cover, nd there re lot of imge overlp condition exists, so the trnsmitted dt the mount cn be effectively reduced, the trnsmission power consumption cn be gretly sved, but it wl significntly increse the dt in the clcultion of energy consumption, nd most clcultions re repeted. Bsed on the literture [], bsed on the use of utomtic imge stitching method to describe the sptil correltion of the imge, nd by mens of utomtic imge stitching method for imge dt redundncy gretly reduced, so tht the mount of dt trnsmission is gretly the reduction, the trnsmission power is reduced; on this bsis, proposed new method of clculting the object of this method of clcultion of the correltion coefficient of the imge. The specific method is the cse, defined the imge pixel s ( i, Y i ), which is collected cmer nodes, tht the number of it is. Becuse the cmer sensor network nodes use the sme terms in the model, without loss of generlity, then we cn set those imges resolution s rows nd N columns [3]. After spce conversion, the originl coordintes ( i, Y i ) wl be chnged to ( y, Y y ). Number i imge spce trnsformtion mtrix is defined s T ik, k hs vlue of, 2, 3, 4, then fter trnsformtion, new pixel coordintes ( y, Y y ) cn be expressed s: 4 i Y y = Y i () k = In this formul, it is possible to drw trnsltion trnsform, respectively, rottion trnsformtion, scling trnsform nd sher trnsformtion mtrix, respectively with T i, T i2, T i3 nd T i4 to represent [4]. 6. ethods Description Rndom dense deployments cmer nodes tht do not optimize it, then there must be certin degree of redundnt nodes by the cmer imge dt were collected. So in order for the cmer to reduce the energy consumption of node communiction, very effective wy is to eliminte redundnt dt exists in this prt, to ensure tht the next hop node trnsmission, wl not hve this prt of the redundnt dt. Another proposed in the literture to remove redundnt dt, this method proposed is bsed on 2D cmer node coverge model. This method is specificlly referring to is the use of the cmer's coverge re overlps the node position, combined with its reltions perspective, wl this type of cmer imges collected node redundnt prt to find out, so tht the cmer nodes only prt of the dt independently trnsmission, the djcent bse sttion node imges cn be spliced [5]. In this wy, if the cmer is t the coverge re of the node is the sme horizontl plne, then the trnsmission of redundnt dt cn be gretly reduced, however, we need to note tht, due to the vrious spects of the constrints, the ctul imge sensor networks re often unble to meet these requirements [6]. In this pper, hypothesized network topology, the topology is divided into clusters; s shown in Fig. : Fig.. Network topology of cluster. In the network topology, the cmer node cluster hed nodes to the cluster mngement, computing cpbity of cluster hed nodes meet the relevnt requirements, nd the storge spce nd bttery cpcity. We use the technique of suture to hndle the cmer node imges within the cluster, nd ssuming tht the network hs been in ccordnce with the clustering method to cluster ws good right division. Bsed on this ide, we put forwrd new method: in the beginning of the work, the imge to the cluster hed node sends the cmer node in this cluster re collected; the cmer imge processing nodes is ccomplished by cluster hed node. It contins these contents. One is to chnge the first cquisition or cmer node position nd ngle, the imge then cn be combined with the reference to the method of imge stitching method of three steps to hndle 3

4 cmer node to collect [7], nd in the process sving process, sptil vrition of the imge mtrix. Two without chnging the cmer node position nd ngle, then in the imge trnsformtion nd splicing of cmer node collection, only need to use the originl prmeters cn be clculted. On this bsis, the redundnt dt on cmer node trnsmission to reduce. After imge stitching, we wl set ech sub grph prmeter s U it, V it, it, Y it, which is in corresponding to U i, V i, i, Y i, so prmeter is U i =u, V i =v, i = R i, Y i = Y R i. Need to py ttention to is, we tlked bout here t U, V,, Y number of lines nd the pixel position, re reltive, reltive to reference position of ech sub grph itself, the reference position new imge it is not generted fter the suture. Imge stitching is good, then the corresponding to ech prt of the originl imge is clled sub grph. We denoted the scling fctor of I imge stitching process s Ri [8]. The cluster hed node send nd Y to the number I cmer node, so in the next cquisition, cmer node only needs to combine the vlues nd Y, nd from their pixel send position to the cluster hed which collected from OS imge sensors; the cmer to collect the cluster hed node, the imge hs no trnsmission of prtilly fled, then rectngulr imge reconstruction [9], nd send the cmer node, so spce trnsform the next nd splicing work wl become esier. Through selected by the imge to the bse sttion, nodes send primry suture of the cluster hed node, the bse sttion node to pply imge pnorm strengthen, gin djustment to primry suture, imge bndwidth rnge nd to further optimize the imge fter primry suture, the visul effect of incresed. 7. lcultion ethod for orreltion oefficient of Imge If the cmer nodes re densely deployed, the coverge re overlp phenomenon is bound to exist, cn be mesured by the correltion coefficient the degree of overlp of these regions. In the references given in the clcultions to cover the cmer node 2D model to clculte the correltion coefficient nd other references re lso given in the method of clcultion, in order to cover the cmer model 3D node correltion coefficient clcultion. Both methods, however there re significnt limittions tht these two clcultion methods re bsed on sptil loction of the cmer, tht cmer node position nd perspective is necessry to understnd, so tht the flexibity of the cmer node deployment wl be gretly limited role in reducing the extent of its fult tolernce, nd requires the cmer hs sensor node, the only wy to tke their own position nd ngle of perception, but the node cost nd power consumption hs been gretly node increse [0]. The proposed method of clculting the ctul fce of the cmer imge cquisition correltion coefficient, by this method my be tht the correltion coefficient bsed on the sptil loction of the node clcultion of the problems solved effectively. We wl defined number i nd j imge orreltion oefficient s P, nd so s to stisfy P = / S, P nd meet between 0 nd. The overlpping re of pictures I nd imge J is expressed s, nd the totl re of pictures I nd imge J fter spliced is expressed s S. The number of pixels used to represent the re. Then P cn be clculted using the following method:. If there is bonded reltionship exists between the i nd j, then turn to 2 directly, or cn mke P equls to If the imge from left to splice j nd i imge, then you cn come to such formul: P i l i = i = = j S * N * 2 i = (2) 3. If the imge from right to splice j nd i imge, then you cn come to such formul: P I = = = S * N * 2 ( N ) l = ( N ) (3) Wht combintion of these imges cn be defined in the correltion coefficient with correltion coefficient of two imges to led out. We mke such definition, σ= /S, the collection of imges to exmine represented by, ll the overlpping regions in imge is expressed s, totl imge re of ll individul imges in imge is expressed s S, the imge re which does not overlp ech other seprte prt in picture is expressed s D, then you cn come to such formul: n N nn + D D nn D δ = =, (4) S nn + D = 8. Simultion n N nn + i i l = L = l = L = ( Y ( Y ) ) (5) ombined with the steps mentioned bove, we toke simultion. We wl focus on co-existence of group of imges s experimentl subjects, s shown below (Fig. 2), these three test imges hve the sme size. 4

5 () (b) (c) Fig. 2. The three experimentl imges After the simultion, we found tht the proposed method hs good dptbity, cn effectively solve the problem of dt redundncy. 9. onclusions This pper propose concept nd clcultion methods bout correltion coefficient of the cmer imge, use the imge stitching technology for wireless multimedi network cmer nodes reduce redundnt dt, the correltion coefficient clculted on the originl issues tht rise from the fundmentl to be solution. The results show tht using the proposed method, the redundnt dt trnsmission cn be effectively reduced, the cmer node communiction consumption hs been reduced. References []. S. Zhenghui, Ruzhung Wng, Hiping Hung, Imge sensor networks bsed on the technique of suture to eliminte redundnt dt, Journl of Southest University, 2, 0, 202, pp [2]. S. Zhenghui, Wireless multimedi sensor network imge stitching technology reserch, Nnjing Post nd ommunictions University, 2,, 202, pp [3]. S. Rong Wu, Wireless multimedi sensor network dt fusion technology reserch, Nnjing Post nd ommunictions University, 5, 2, 995, pp [4]. S. Zheyun iong, Wireless multimedi sensor network imge coding lgorithms, entrl South University, 2, 5, 202, pp [5]. S. Lu To, Zhu Qing-in, Zhu Yu-Yu, An Energy- Efficient Adptive lustering Protocol for Wireless Sensor Network, Sensors & Trnsducers, Vol. 52, y 5, 203, pp [6]. S. Li Wei, Lightweight wireless sensor network security reserch of dt fusion scheme, Nnjing Post nd ommunictions University, 2, 3, 202, pp [7]. S. Zhi Zhong, With mobe nodes of wireless sensor network locliztion lgorithm nd dt collection protocol reserch, Nnjing Post nd ommunictions University, 5, 2, 202, pp [8]. S. Zhiwei Lee, Bsed on the clustering of the wireless sensor network lifetime extension strtegy reserch, Ocen University of hin, 2, 5, 2006, pp [9]. S. Lei Yn, iokng Ding, Zheng Yu, Jinlei Kong, Jinho Liu, A Novel Identifiction ethod of Obstcles Bsed on ulti-sensor Dt Fusion in Forest, Sensors & Trnsducers, Vol. 55, Issue 8, August 203, pp [0]. Hu Ouyng, Hui Li, ei Qin. ompression of Power Qulity Dt Bsed on Improved DT Trnsform, Sensors & Trnsducers, Vol.9, Specil Issue, 203, pp opyright, Interntionl Frequency Sensor Assocition (IFSA) Publishing, S. L. All rights reserved. ( 5