Search

In this article, a multiview image compression framework, which involves the use of Block-based Compressive Sensing (BCS) and Joint Multiphase Decoding (JMD), is proposed for a Visual Sensor Network (VSN). In the proposed framework, one of the sensor nodes is configured to serve as the reference node, the others as nonreference nodes. The images are encoded independently using the BCS to produce two observed measurements that are transmitted to the host workstation. In this case, the nonreference nodes always encoded the images (I NR ) at a lower subrate when compared with the images from the reference nodes (I R ). The idea is to improve the reconstruction of I NR using I R . After the two observed measurements are received by the host workstation, they are first decoded independently, then image registration is applied to align I R onto the same plane of I NR . The aligned I R is then fused with I NR , using wavelets to produce the projected image I P . Subsequently, the difference between the measurements of the I P and I NR is calculated. The difference is then decoded and added to I P to produce the final reconstructed I NR . The simulation results show that the proposed framework is able to improve the quality of I NR on average by 2dB to 3dB at lower subrates when compared with other Compressive Sensing (CS)--based multiview image compression frameworks.

When the embedded zerotree wavelet (EZW) algorithm was first introduced by Shapiro, four types of symbols (zerotree (ZTR), isolated zero (IZ), positive (POS), and negative (NEG)) were used to represent the tree structure. An improved version of EZW, the set partitioning in hierarchical trees (SPIHT) algorithm was later proposed by Said and Pearlman. SPIHT removed the ZTR symbol, while keeping the other three symbols in a slightly different form. In the SPIHT algorithm, the coding of the parent node is isolated from the coding of its descendants in the tree structure. Therefore, it is no longer possible to encode the parent and its descendants with a single symbol. When both the parent and its descendants are insignificant (forming a degree-0 zerotree (ZTR)), it cannot be represented using a ZTR symbol. From our observation, the number of degree-0 ZTRs can occur very frequently not only in natural and synthesis images, but also in video sequences. Hence, the ZTR symbol is reintroduced into SPIHT in our proposed SPIHT-ZTR algorithm. In order to achieve this, the order of sending the output bits was modified to accommodate the use of ZTR symbol. Moreover, the significant offspring were also encoded using a slightly different method to further enhance the performance. The SPIHT-ZTR algorithm was evaluated on images and video sequences. From the simulation results, the performance of binary-uncoded SPIHT-ZTR is higher than binary-uncoded SPIHT and close to SPIHT with adaptive arithmetic coding.

This paper proposed an energy efficient multi-view visual processing scheme for the visual sensor networks (VSNs). The aim of this scheme is to reduce the energy consumption by reducing the amount of data need to be transmitted over the network. This is achieved by using two approaches. First, an image compression based on the Set Partitioning in Hierarchical Trees (SPIHT) [1] algorithm is implemented to compress the images before transmitting them over the network. Second, in cases where the Field-of-View (FoV) of different visual nodes is overlapped, the overlapping regions (redundant information) between the images will be identified at the host workstation to prevent them from being transmitted multiple times. This is to further reduce the data transmission. In this case, only one visual node is required to transmit the overlapping regions. The information obtained from the aforementioned visual node will be used to reconstruct the overlapping regions of other visual nodes at the host workstation. In other words, other visual nodes are only required to transmit the non-overlapping regions. Instead of transmitting the entire image over the network, only part of it will be transmitted. The simulations results show that our proposed scheme can reduce the data transmission for 50%. Furthermore, the quality of reconstructed images is acceptable and can be improved in further works.

A multiview image compression framework for the Wireless Multimedia Sensor Network (WMSN) based on image stitching and SPIHT coding is proposed. This framework is designed to work with the mesh network topology that is very robust. In this framework, the images taken by neighboring sensors are stitched together with an image stitching technique to remove the overlap redundancy. The image stitching process will be carried out in certain intermediate nodes along the way towards a centralized decoder. This will help to conserve more power since the amount of data to be transmitted is reduced. In addition, the tree structure adapted by SPIHT is modified to improve the flexibility in coding the stitched image. Simulation results show that the use of image stitching with SPIHT coding can greatly remove the overlap and spatial redundancy.

An image compression technique that uses the Harris corner detector and SPIHT coding to encode images from two cameras with overlapping field of view is proposed in this paper. Firstly, the Harris corner detector is used to extract the feature points in the two images separately. Secondly, the feature points are matched and the homography that relates the two images is estimated using RANSAC. Finally, the two images are stitched together and the SPIHT coding is used to compress the stitched image. Since the resolution of the stitched image is not always fixed and difficult to be predetermined, the tree structure that is adopted by the SPIHT coding is slightly modified to resolve the problem. Generally, this technique can also be viewed as an alternative solution for joint sources image compression. The simulation results show that the file size can be reduced by approximately 14.25-25.49% depending on the amount of the overlapping region.

Due to the limited Field-Of-View (FOV) of a single camera, it is sometimes desired to extend the FOV using multiple cameras. Image stitching is one of the methods that can be used to exploit and remove the redundancy created by the overlapping FOV. However, the memory requirement and the amount of computation for conventional implementation of image stitching are very high. In this paper, this problem is resolved by performing the image stitching and compression in a strip-by-strip manner. First, the stitching parameters are determined by transmitting two reference images to an intermediate node to perform the processing. Then, these parameters are transmitted back to the visual node and stored in there. These parameters will be used to determine the way of stitching the incoming images in a strip-by-strip manner. After the stitching of a strip is done, it can be further compressed using a strip-based compression technique.