High-efficiency Coding And Transmission Method In WISNs For Wild Monitoring Images

Wireless image sensor networks(WISNs)have broad development potential in the field of remote,real-time and accurate information monitoring.Aiming at the challenges of monitoring sensor node such as the strong power consumption limitations,complex perception environment,low reconstruction quality of important region,slow transmission speed and weak anti-interference ability,it is the key for solving the problem to explore the high-efficiency coding and transmission method for large data amounts and strong noise images.Taking the wildlife monitoring as an application scenario,this paper aims to balance the reconstruction quality,transmission efficiency and power consumption of wildlife monitoring images in the WISNs mode,and strives to achieve efficient coding and transmission of wildlife images.First of all,the saliency object detection algorithm based on improved histogram contrast is proposed in this paper,and then the mask image of saliency object region is generated,which can be used to raise the transmission priority of the pixels in the important region;Secondly,an efficient transmission strategy of image progressive compression coding and image data assign under the distributed mechanism is explored,which ensure the reconstruction quality of important region and allocate the network resources reasonably.Furthermore,an automatic restoration algorithm of missing image based on auto-encoder structure is proposed to improve the availability of the image sample under the complex conditions,which can provide the data guarantee for subsequent scientific research.(1)Aiming at the problem of data acquisition lagging in wild environment,a remote monitoring system based on WISNs was designed and implemented,and the image sample library of monitoring images is established.A total of 10720 wildlife image sample libraries have been established,which inclued six species wildlife of Red deer,Chinese goral,Roe deer,Lynx,Wild boar and Raccoon dog.Besides,the ground truth images are also produced,which provide theoretical basis for subsequent experimental comparisons of saliency object detection,image compression and transmission algorithms.(2)Aiming at the problem of monitoring images with complex background,large amount data and serious noise interference,an improved image saliency object detection algorithm based on histogram contrast is explored.Based on the traditional histogram contrast algorithm,this paper combines the strategies of structure extraction and position saliency map to achieve the the detection and extraction of image saliency object region,which smooths the image texture information and suppresses the noise interference.The average Pr,Re and F-measure values of experiments conducted on the wildlifemonitoring sample have reached 0.4895,0.7321 and 0.5300 respectively.Compared with the HC and MC algorithms,they were increased by 18.37%,19.53%,19.05% and 6.42%,21.99%,8.74%.(3)In order to solve the problem that traditional image compression algorithms can not reflect the priority of important region,an image progressive compression coding algorithm was proposed based on saliency visual perception.This paper,on the basis of image salient target detection result,utilizes the bit plane raising and hybrid coding algorithm to perform layered and progressive compression coding on wildlife images,respectively achieving lossless compression of saliency object region and lossy compression of background region to ensure the reconstruction quality the important region information.The average results of PSNR and SSIM of our algorithm are 39.0365 d B and 0.9014,which are improved by 21.11%,14.72% and 9.47%,6.25% respectively when compared with EZW and DCT.(4)Aiming at the problem of network resource waste caused by the self-organizing and multi-hop transmission mode of WISNs,an image data distribution strategy based on distributed transmission mechanism was explored in this paper.A distributed cooperative transmission strategy of salient object region and background region is proposed by independent coding and joint decoding for joint signals.In the algorithm,the sailency object region is directly transmitted by cluster head nodes,and the background region whose data is relatively large is reasonably distributed among the nodes in the cluster at the same transmission level,which realizes the rational use of network resources.Compared with DCT and EZW,the proposed algorithm has improved by 7.47%,9.06% and 16.98%,19.50%respectively.In terms of energy consumption,our algorithm is reduced by 29.96% and 40.84%compared with the single transmission mode and multi-hop transmission mode.(5)To solve the problem of image content mission in WISNs caused by external environment interference,the automatic image restoration algorithm based on improved auto-encoder is proposed.Aiming at the difference in texture information between saliency object region and background region,an improved network structure based on improved auto-encoder is proposed by training and testing the sample images of the saliency object region and the background region separately.It realizes the automatic recovery of important missing information in the image samples,and the experimental results show that the PSNR and SSIM are improved by 7.93%,18.15% and 7.01%,12.67% respectively,which ensures the reliability of monitoring data and provides the materials for subsequent related scientific research.In summary,this paper proposes an efficient coding and transmission method in WISNs for large-amount data and complex background wild monitoring images,including the design of WISNs monitoring system,image saliency object detection,image progressive compression coding and the distributed transmission strategy,the automatic restoration of missing images,which can provide the theoretical guidance for the promotion of WISNs in the field of intelligent information monitoring.