Institutional Repository of Xi'an Institute of Optics and Precision Mechanics, CAS
Not a member yet
30565 research outputs found
Sort by
Exploiting spatial relation for fine-grained image classification
No full textFine-Grained Image Classification (FGIC) aims to distinguish the images within a subordinate category. Recently, many FGIC methods have been proposed and huge progress has been made in the aspects of part detection and feature learning for FGIC. However, FGIC still remains a challenging task due to the large intra-class variance and small inter-class variance. To classify fine-grained images accurately, this paper proposes to exploit spatial relation to capture more discriminative details for FGIC. The proposed method contains two core modules: part selection module and representation module. The part selection module utilizes intrinsic spatial relation between object parts to select object part pairs with high discrimination power. The representation module exploits the interaction between object parts to describe the selected part pairs and construct a semantic image representation for FGIC. The proposed method is evaluated on CUB-200-2011 and FGVC-Aircraft datasets. Experimental results show that the classification accuracy of the proposed method can reach 85.5% on CUB-200-2011 and 86.9% on FGVC-Aircraft respectively, which exceed comparison methods obviously. ? 2019 Elsevier Ltd</p
Text recognition from silent lip movement video
No full textLip reading, the ability to recognize text information from the movement of a speaker's mouth, is a difficult and challenging task. Recently, the end-to-end model that maps a variable-length sequence of video frames to text performs poorly in real life situation where people unintentionally move the lips instead of speaking. The goal of this work is to improve the performance of lip reading task in real life. The model proposed in this article consists of two networks that are visual to audio feature network and audio feature to text network. Our experiments showed that the model proposed in this article can achieve 92.76% accuracy in lip reading task on the dataset that the unintentional lips movement was added. ? 2018 IEEE
Design of deviation detector for image stabilization system on space solar observation equipment
No full textA moving target extraction algorithm based on the fusion of infrared and visible images
No full textAccording to the principle of thermal imaging, moving targets can be better located in infrared images, but their boundary is blurred, and the details of objects cannot be displayed. The details of objects in natural images can be better shown, but for the condition of shelter, shadow and etc., miss-tracking and false-tracking may easily occur. Thus, we construct a framework for moving target extraction and tracking in infrared and natural images. For infrared images: according to the rough fuzzy set theory, we propose the rough entropy model based on the traditional frame difference method. The model is fused with the infrared imaging characteristics to locate moving target regions. For natural images: a time-space fusion LBP model is proposed for target coding. The model is integrated into the GMM model to obtain moving target information. The moving regions in infrared images and natural images are fused to form a priori model, and the C-V model is improved to extract targets accurately. © 2019</p
Observation of evolution dynamics from bound states to single-pulse states in a passively mode-locked fiber laser
No full textWe report what is, to our knowledge, the first experimental observation of the ultrafast evolution dynamics from bound states (BSs) to single-pulse states (SPSs) by using the dispersive Fourier-transform (DFT) technique. The evolutions from three categories of initial BSs to SPSs are spectrally resolved in real time. Usually, accompanied by complex soliton-soliton interaction and competition, one of the two bound pulses weakens to disappearance, and the other one evolves into SPS. During the transition, the two bound pulses ordinarily depart away from each other with complex changes of relative phase. However, it is found that not all the evolutions are accompanied by the increase of temporal separation between two bound pulses. The obtained results would facilitate a deep understanding of complex dynamics in nonlinear systems and provide valuable data for further theoretical studies.</p
Linear space-variant optical cryptosystem via Fourier ptychography
No full textAn optical cryptosystem via Fourier ptychography with double random phase masks is proposed. The encryption process cannot be precisely simulated except by optical experiment due to vignetting, which is space variant and can act as a one-way function from the perspective of pure optics and improve the security of our system. In addition, the encryption for a high-resolution, large field-of-view, and complex-valued image is achievable. Optical experiments are presented to prove the validity and security of the proposed system. Our method will give more insight into separating optical cryptography from computer cryptography in nature. (C) 2019 Optical Society of America</p
具有同步控制功能的点源透过率杂光测试系统及方法
No full text本发明公开一种具有同步控制功能的点源透过率杂光测试系统及方法，属于杂光测试领域。光源出射的脉冲激光经平行光管准直后出射平行光，一部分光直接进入待测光机系统，该部分光为信号光；其中一部分光经待测光机系统表面散射后照亮环境内壁，再由内壁散射返回待测光机系统，该部分光为环境光污染。相对于信号光，环境光污染须经历更长的光程和时间到达系统像面，具有一定的时间滞后，当信号光达到系统焦面时，同步控制系统控制探测器快门打开，接收信号光；而当环境光污染到达系统焦面时，快门处于关闭状态。本发明可有效抑制环境光污染，提高点源透过率杂光测试系统的精度。</p
