CN102340667A - Distributed image transmission method oriented to wireless multimedia sensor network - Google Patents

Abstract

The invention relates to a distributed image transmission method oriented to a wireless multimedia sensor network, which is designed to solve the problem that limited resources cannot be fully utilized because the resources of the wireless multimedia sensor network are seriously limited and the energy consumption stresses of nodes are not balanced. The method takes a concurrent computation concept as a basis and is based on the high-efficiency energy-saving distributed image compression algorithm of JPEG2000. During clustering, the algorithm introduces a node campaign factor and a communication cost function, self-adaptive optimal distribution of compression bit rates is conducted to an image block according to the gradient magnitude of an image and finally data information molecule bands after wavelet transformation in an image compression process are separately coded. The distributed image transmission method oriented to the wireless multimedia sensor network has the characteristics and the advantages that on the premise that image quality is guaranteed, the large-size high-resolution image transmission is realized in the wireless multimedia sensor network with seriously limited resources at low energy consumption and in load balance, the energy consumption stresses of camera nodes can be greatly relieved, the image acquisition and compression energy consumption stresses of a cluster-head node layer can be decomposed to a redundantly-deployed common node layer through the cooperation between the common node layers and the cluster-head node layer, the energy consumed by the cluster-head node layer can be effectively decreased by 25 percent in each turn, at the same time, the energy consumption of network nodes can be greatly balanced and the lifecycle of the network is prolonged by 15 percent.

Description

A kind of distributed image transfer method towards wireless multimedia sensor network

Technical field

The present invention relates to a kind of image transfer method, relate in particular to a kind of distributed image transfer method towards wireless multimedia sensor network.

Background technology

Along with the development of hardware technology, the cheap CMOS camera and the appearance of microphone, and, caused the appearance of wireless multimedia sensor network (WMSNs) to the image that contains abundant information of monitoring target, the active demand of Voice & Video data.The distributed sensing network that WMSNs is made up of one group of multimedia sensor node with calculating, storage and communication capacity generally is deployed in the autonomous task of accomplishing appointment in the unattended environment, is the responsive foundation-free facility network of a kind of energy consumption.Typical WMSNs generally is made up of multimedia sensor node, aggregation node, base station; It belongs to multiple medium (audio frequency, video, image, the numerical value etc.) information of surrounding enviroment by means of multi-media sensor perception on the node, thereby is sent to aggregation node through the multi-hop route with wireless mode then. and autonomous realization is to the monitoring of surrounding environment.

IMAQ is to carry out two the most key technology of image transmission at WMSNs with compression, directly has influence on the service quality of its perception.To in WMSNs, carry out the collection and the transmission of image; Node in the network must satisfy the real-time collection to initial data; The compression ratio of node requirement simultaneously is higher; Existing traditional image compression algorithm calculation of complex, expense are big, and are not suitable for the serious limited wireless multimedia sensor of resource.In addition, it is bigger to comprise data of image information wireless receiving and dispatching energy consumption, and how these data also are problem demanding prompt solutions in the transmission of network low energy consumption.

WMSNs combines sensor technology, multimedia technology, embedded computing technique, distributed information processing, wireless communication technology etc., is the research field that novel a, forward position, multidisciplinary height intersect.Compare with the traditional wireless sensor networks that only has simple environmental data collection function; Wireless multimedia sensor network expands to physical space arbitrarily with the mankind's the visual field; Changed the interactive mode of people and physical world deeply, had boundless application prospect in fields such as military, civilian and commerce.Because the huge using value of WMSNs, it has caused the very big concern of world many countries military service and academia.In time carry out the research that this has the frontier science and technology of profound influence, obtain to have the correlation technique of China's independent intellectual property right, be of great immediate significance and far-reaching strategic importance.

Summary of the invention

Because the wireless multimedia sensor network resource is seriously limited, and each node energy consumption pressure is unbalanced, the problem that limited resources are not fully utilized in order to solve in the present invention; And according to the characteristics of WMSNs topological structure; A kind of distributed image transfer method towards wireless multimedia sensor network is provided, and this method is the basis with parallel computation thought, based on the distributed image transmission of JPEG2000; Employing is divided into the structure of ordinary node layer and camera node layer with node, and its step is following:

Step 1: wake camera node up, when the node of ordinary node layer detected the monitored area and has abnormal conditions to occur, their can wake camera node up by agreement mechanism;

Step 2: sub-clustering, select bunch head to introduce the node election contest factor and a communication cost function, self-organizing formation bunch;

Step 3: image processing, the synthetic image visual signature carries out distributed image compression and transmission.

Characteristics of the present invention and beneficial effect: under the situation that guarantees picture quality; With low energy consumption in the serious limited wireless multimedia sensor network of resource; The realization large scale of load balancing, high-resolution image transmit, and can greatly alleviate the energy consumption pressure of camera node, through the cooperation of ordinary node layer and leader cluster node interlayer; The IMAQ of leader cluster node layer is decomposed the ordinary node layer of disposing redundancy with compression energy consumption pressure; Every round effectively reduces leader cluster node layer 25% energy consumption, and the while is the balancing network node energy consumption greatly, prolongs 15% network lifecycle.

Description of drawings:

Fig. 1 camera node wakes flow chart up;

Fig. 2 IMAQ and transfer process figure;

Fig. 3 JPEG2000 distributed coding block diagram.

Embodiment:

Referring to Fig. 1-Fig. 3; A kind of distributed image transfer method towards wireless multimedia sensor network, this method is the basis with parallel computation thought, based on the distributed image transmission of JPEG2000; Employing is divided into the structure of ordinary node layer and camera node layer with node, and its step is following:

Step 1: wake camera node up, when the node of ordinary node layer detected the monitored area and has abnormal conditions to occur, their can wake camera node up by agreement mechanism;

Step 2: sub-clustering, select bunch head to introduce the node election contest factor and a communication cost function, self-organizing formation bunch;

Step 3: image processing, the synthetic image visual signature carries out distributed image compression and transmission.

Wherein:

It is following that camera node wakes flow process up:

(1) after wireless multimedia sensor network was disposed well, the camera node layer was in resting state, and ordinary node is surveyed surrounding enviroment incessantly;

(2) when Event triggered, the ordinary node reported event is extremely apart from self nearest camera node, and camera node is broadcasted this event message with the communication radius of setting;

(3) competition wakes up the camera node of all these incidents of covering based on self rest energy;

(4) camera node of competition success is broadcasted oneself state message immediately, and the preparation for acquiring image.

The idiographic flow of sub-clustering is following:

(1) base station broadcast is created routing iinformation, and node obtains the election contest factor of oneself based on routing iinformation and dump energy, and campaigns for bunch head;

(2) node of election contest failure is selected bunch head according to the communication cost function.

The idiographic flow of image processing is following:

(1) with the original image piecemeal, define the gradient amplitude value of the marginal information that is comprised in each image block, with its value compression bit rate is carried out adaptive optimization and distribute, piecemeal carries out wavelet transformation;

(2) consider serious limited this problem of WMSNs resource, LL, LH, HL and HH information after adopting distributed coded system to wavelet transformation are carried out distributed EBC coding.

Transfer of data communications protocol in this method is following:

1). camera node and neighbours' leader cluster node adopt the F1 frequency, realize communication with the TDMA mode;

2). all leader cluster nodes and its child node are adopted the F2 frequency, realize communication with the TDMA mode;

3). adopt the CDMA mode between bunch.

Embodiment

The present invention is divided into ordinary node layer and camera node layer with node; After wireless multimedia sensor network is disposed well; The camera node layer generally is in resting state; Abnormal conditions (for example temperature anomaly, sharp pounding etc. is arranged) are arranged when occurring, they can wake camera node up when the node of ordinary node layer detects the monitored area.Camera node is opened camera and is carried out information gathering and carry out preliminary treatment; With the original image piecemeal; Every compression duty is assigned in a plurality of bunches; When sub-clustering, introduce the node election contest factor and communication cost function, the load of node in the balancing network reduces the energy consumption expense of view data in transmission course.The adaptive optimization that carries out compression bit rate according to the gradient amplitude value of image in when compression distributes; Comprise the more image block of marginal information and will obtain more compression bit rate and divide frequency band to encode separately the data message behind the wavelet transformation at last, alleviate the energy consumption pressure of leader cluster node.

With reference to Fig. 1, the camera node arouse machine processed, though the distributed arouse machine processed of the present invention's design needs the camera node participation and requires the expenditure of energy, its communication delay is lower, has good expandability, is more suitable in extensive WMSNs, idiographic flow is following:

Step 1: camera node broadcasting HELLO message, all ordinary nodes write down apart from self nearest camera node as the reporting events node;

Step 2: ordinary node is surveyed surrounding enviroment incessantly.When detecting the monitored area and have abnormal conditions (for example temperature anomaly, sharp pounding etc. is arranged) incident or previous perception events to finish, ordinary node reports to nearest camera node with incident relevant information, self identification, position and the perception radius mode through the minimum hop count route;

Step 3: receive camera node this event message of communication radius broadcasting to set from the ordinary node event message.Then, all camera nodes that receive event message judge according to the ordinary node position and the perception radius of this incident of perception whether self covers this incident.

Step 4: competition wakes up the camera node of all these incidents of covering based on self rest energy; The camera node of competition success is broadcasted oneself state message immediately, and the preparation for acquiring image.

With reference to Fig. 2, the network topology structure of WMSNs when carrying out IMAQ with transmission, mark S is a camera node among the figure; D is an aggregation node; P11-P14 is a leader cluster node; HH _i, HL _i, LH _iAnd LL _iIt is the sub-band information of node processing.

Arbitrary node in network all has the also dump energy of ability perception oneself unique ID number, and camera node and ordinary node are disposed the position, back and all no longer changed, and the practical implementation step of IMAQ and transmission is following:

Step 1: camera node collects informing base station behind the image, and l is broadcasted immediately in the base station in its effective communication radius ₀=0 information, l _iIt is the jumping figure that each node is arrived in the base station.Node is received after this information the l with oneself _iChange l into _i=l ₀+ 1, node i continues broadcasting establishing route information then.When next hop neighbor node of node i is received this information, own jumping figure with the base station is made as l _j=l _i+ 1.Work as l _jDuring greater than preset jumping figure threshold value, then go off the air.If node repeatedly receives establishing route information, then choose the jumping figure value of minimum wherein, the jumping figure of all node bases and base station, computational logic position weights

l _MaxIt is the maximum hop count of node and base station in the network;

Step 2: all nodes are according to the distance of self and camera node, according to formula

Calculate the logical place weights of self, wherein, camera node is s, d _MaxAnd d _MinBe respectively in the network arbitrary node i apart from the ultimate range and the minimum range of camera node.Node i to the distance of camera node be d (i, s).

Step 3: according to formula

Computing network every average energy consumption of taking turns that is in operation, E _{Ave_r}Be every average energy consumption of taking turns node in the network, E _Current(i) be the current remaining of node i, for reaching E _Current(i)>=E _{Ave_r}The node that requires, the leader cluster node of will having an opportunity to compete;

Step 4: the node that reaches the election contest condition is according to formula W (i)=(1-k) * w ₁+ k*w ₂And formula

Calculate oneself apart from the factor and load factor, at last according to formula $T\left(i\right)=\left\{\begin{array}{cc}\frac{p}{1-p\left[r\mathrm{Mod}(1/p)\right]}[\mathrm{\μ}*W\left(i\right)+E_{\mathrm{Foreoast}}\left(i\right)],& \∀i\∈G^{\′}\\ 0,& \∀i\∉G^{\′}\end{array}\right.$ Elect leader cluster node;

Step 5: the distance of leader cluster node basis and camera node, self-organizing forms a virtual cluster chain, and a bunch chain default camera node is first leader cluster node.Remaining ordinary node is according to formula cost (c _i, i)=f ₁(d (c _i, i))+f ₂(d (c _{I_next}, i)), wherein

I is an ordinary node, c _iBe any leader cluster node of electing, c _{I_next}Be leader cluster node c _iNext jump leader cluster node.D (c _i, i) and d (c _{I_next}, be that node i arrives leader cluster node c i) _iAnd c _{I_next}Distance, d _{I_max}Be the ultimate range of node i to the leader cluster node of electing.For arbitrary node i, select cost (c _i, i) the leader cluster node c of minimum _iAdd, self-organizing forms bunch;

Step 6: camera node carries out preliminary treatment to image, and view data is divided into the M piece.Visual signature in conjunction with human eye; Define the gradient amplitude value of the marginal information that is comprised in each image block according to formula

; Through formula g (x; Y)=| f (x; Y)-f (x+1, y+1) |+| f (x+1, y)-f (x; Y+1) | characterize the gradient amplitude value of each image slices vegetarian refreshments, the compression bit rate that calculates according to formula

at last;

Step 7: br by formula _i=w _i* M*Br carries out the adaptive optimization distribution of compression bit rate;

Step 8: camera node is chosen M maximum leader cluster node of dump energy, carries out the data processing task of wavelet transformation;

Step 9: leader cluster node is with 16 final wavelet conversion coefficients of short type storage, and bunch in choose the maximum node of dump energy LL, LH, HL and HH information encoded separately;

Step 10: if do not meet the requirements of compression ratio, then LL information is gathered to camera node, according to step 5, carry out the next round image processing, all the other information behind the coding, transfer to the base station according to step 8;

Step 11: so go on, the data after will encoding at last will be by leader cluster node with code check br _iAdopt the multi-hop communication mode to transfer to the base station.

In addition, for wireless multimedia sensor network, the jumping figure between source node and the destination node is enough big for single bunch of one-level small echo, so, near the node of aggregation node too big calculated load is arranged unlikely.

With reference to the JPEG2000 coding block diagram after Fig. 3 improvement, the present invention is according to the concrete characteristics of WMSNs network, and this paper improves original, centralized JPEG2000 encryption algorithm from three aspects.

Step 1: to the storage of WMSNs individual node, the limited problem of computing capability; This programme carries out preliminary treatment through camera node to raw image data; Raw image data is divided into piece, piecemeal is replaced traditional wavelet transform with distributed wavelet transform;

Step 2: quantize, ins conjunction with visual characteristics of human eyes, comprise the information at edge according to image block, adaptive optimization distributes compression bit rate, with assurance picture quality;

Step 3: consider serious limited this problem of WMSNs resource, make full use of and dispose redundant ordinary node layer, adopt distributed coded system that LL, LH, HL and HH information are carried out distributed EBC coding;

Step 4: when not meeting the requirements of compression ratio; The LL information reconstructed picture of sensitivity when only human eye being discerned; Resultant four LL sub-band informations are carried out the secondary wavelet transformation; The information of all the other frequency range LH, HL and HH is then directly pressed the distributed EBC coding of subband, forms the JPEG2000 code stream and transfers to aggregation node D.

Claims (5)

1. distributed image transfer method towards wireless multimedia sensor network; This method is the basis with parallel computation thought; Based on the distributed image transmission of JPEG2000, adopt the structure that node is divided into ordinary node layer and camera node layer, its step is following:

Step 1: wake camera node up, when the node of ordinary node layer detected the monitored area and has abnormal conditions to occur, their can wake camera node up by agreement mechanism;

Step 2: sub-clustering, select bunch head to introduce the node election contest factor and a communication cost function, self-organizing formation bunch;

Step 3: image processing, the synthetic image visual signature carries out distributed image compression and transmission.

2. require described a kind of distributed image transfer method towards wireless multimedia sensor network according to right 1, it is characterized in that: it is following that said camera node wakes flow process up:

(1) after wireless multimedia sensor network was disposed well, the camera node layer was in resting state, and ordinary node is surveyed surrounding enviroment incessantly;

(2) when Event triggered, the ordinary node reported event is extremely apart from self nearest camera node, and camera node is broadcasted this event message with the communication radius of setting;

(3) competition wakes up the camera node of all these incidents of covering based on self rest energy;

(4) camera node of competition success is broadcasted oneself state message immediately, and the preparation for acquiring image.

3. require described a kind of distributed image transfer method towards wireless multimedia sensor network according to right 1, it is characterized in that: the flow process of said sub-clustering is following:

(1) base station broadcast is created routing iinformation, and node obtains the election contest factor of oneself based on routing iinformation and dump energy, and campaigns for bunch head;

(2) node of election contest failure is selected bunch head according to the communication cost function.

4. require described a kind of distributed image transfer method towards wireless multimedia sensor network according to right 1, it is characterized in that: the flow process of said image processing is following:

(1) with the original image piecemeal, define the gradient amplitude value of the marginal information that is comprised in each image block, with its value compression bit rate is carried out adaptive optimization and distribute, piecemeal carries out wavelet transformation;

(2) consider serious limited this problem of WMSNs resource, LL, LH, HL and HH information after adopting distributed coded system to wavelet transformation are carried out distributed EBC coding.

5. require described a kind of distributed image transfer method towards wireless multimedia sensor network according to right 1, it is characterized in that: the communications protocol of transfer of data is following in this method:

1). camera node and neighbours' leader cluster node adopt the F1 frequency, realize communication with the TDMA mode;

2). all leader cluster nodes and its child node are adopted the F2 frequency, realize communication with the TDMA mode;

3). adopt the CDMA mode between bunch.

Images