Recently, convolutional neural networks (CNNs) achieve impressive results on remote sensing scene classification, which is a fundamental problem for scene semantic understanding. However, convolution, the most essential operation in CNNs, restricts the development of CNN-based methods for scene classification. Convolution is not efficient enough for high-resolution remote sensing images and limited in extracting discriminative features due to its linearity. Thus, there has been growing interest in improving the convolutional layer. The hardware implementation of the JPEG2000 standard relies on the lifting scheme to perform wavelet transform (WT). Compared with the convolution-based two-channel filter bank method of WT, the lifting scheme is faster, taking up less storage and having the ability of nonlinear transformation. Therefore, the lifting scheme can be regarded as a better alternative implementation for convolution in vanilla CNNs. This paper introduces the lifting scheme into deep learning and addresses the problems that only fixed and finite wavelet bases can be replaced by the lifting scheme, and the parameters cannot be updated through backpropagation. This paper proves that any convolutional layer in vanilla CNNs can be substituted by an equivalent lifting scheme. A lifting scheme-based deep neural network (LSNet) is presented to promote network applications on computational-limited platforms and utilize the nonlinearity of the lifting scheme to enhance performance. LSNet is validated on the CIFAR-100 dataset and the overall accuracies increase by 2.48% and 1.38% in the 1D and 2D experiments respectively. Experimental results on the AID which is one of the newest remote sensing scene dataset demonstrate that 1D LSNet and 2D LSNet achieve 2.05% and 0.45% accuracy improvement compared with the vanilla CNNs respectively.
A novel digital watermarking scheme is proposed instead of the traditional watermarking.The binary watermark image is processed by error-correcting coding before it is embedded into the discrete wavelet-transformed host image.It is more effective to improve the security and robustness of the watermarked image by proposed scheme.Amount tests and results indicate that the proposed scheme have security and robustness against noise and commonly image processing methods such as Gaussian's noise, filtering, JPEG compression, and crop procession etc.
At present, there are no recognized guidelines or consensus for the treatment strategy of the asymptomatic tooth with external root resorption caused by an embedded tooth (et-ERR). Most clinicians would like prophylactic or concomitant root canal therapy (RCT) along with the extraction of the embedded tooth. The purpose of this study was to report the prognosis of external root resorption (ERR) and investigate the possibility to preserve the vital pulp of ERR tooth.The patients who had asymptomatic et-ERR teeth were included. After extraction of the embedded tooth, the clinical process, prognosis, and adverse events were observed, including symptoms, clinical, and radiographic examination throughout the follow-up period.A total of four cases with special features were reported. Over a follow-up period of up to 12 months, on clinical examination, 3 ERR teeth preserved pulp vitality without additional intervention except for tooth extraction and have kept normal function free from any symptoms. Radiographic examination showed bone regeneration and recovery of periodontal tissue. While one case failed to keep the vital pulp and ended in intentional replantation.As to et-ERR, if the embedded tooth can be promptly extracted with a minimally invasive technique and effective infection control, the pulp vitality of the et-ERR tooth is likely to be preserved. In this situation, the preferred management of asymptomatic et-ERR tooth is just followed up without prophylactic RCT.
Electrospinning technology has been widely used in the past few decades to prepare nanofibrous scaffolds that mimic extracellular matrices. However, traditional two-dimensional (2D) electrospun nanofibrous mats still have some inherent disadvantages for bone tissue engineering, such as limited cell infiltration and lack of three-dimensional (3D) structure. The development of 3D electrospun scaffolds with larger pore sizes and porosity provides new perspectives for electrospinning-based tissue engineering scaffolds. However, there are still some challenges and areas for improvement. In this review, the applications of 3D electrospun nanofibrous scaffolds in the field of bone tissue engineering from its fabrication methods to bio-functionalization are summarized, with the aim of providing new insights into the design of electrospinning-based bone tissue engineering scaffolds.
Aiming at high fidelity image only can be encrypted in time domain, summarize the current all kind of encryption methods in time domain. That are only chaos permutation, only chaos encryption and hybrid encryption algorithm. Then, this paper proposed the block chaos image scrambling and chaos encryption with after effect. In the block permutation, give the formula of correlation and compare the correlation before and after permutation. After the hybrid encryption, computing the information entropy of encrypted image. At last, according to the security analysis, the image encryption algorithm demonstrates strong resistance toward exterior attacks such as statistical attacks and differential attacks.
Some problems existed in IEEE 1394a camera of the Trouble of Moving Freight Car Detection System (TFDS), such as easy disconnection with electromagnetic interference, limited image resolution and low frame rate. To improve that, a GigE camera is used as a replacement. Images of GigE camera are acquired and processed using embedded system based on FPGA and DSP. First gray images are acquired and compressed by JPEG. Then compressed data is saved on hard disk and uploaded to the server in detecting center simultaneously. System construction is greatly simplified ,meanwhile the costs, volume and power consumption are also decreased. The system has worked with a stable status for a long period in the actual railway. The final tests demonstrate that the design is feasible of great application potential .
A novel iris biometric watermarking scheme is proposed focusing on iris recognition instead of the traditional watermark for increasing the security of the digital products. The preprocess of iris image is to be done firstly, which generates the iris biometric template from person's eye images. And then the templates are to be on discrete cosine transform; the value of the discrete cosine is encoded to BCH error control coding. The host image is divided into four areas equally correspondingly. The BCH codes are embedded in the singular values of each host image's coefficients which are obtained through discrete cosine transform (DCT). Numerical results reveal that proposed method can extract the watermark effectively and illustrate its security and robustness.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)