Figure 3 - available via license: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Content may be subject to copyright.
Source publication
In this paper, an optical correlator which adopts planar integrated 2F structure and optimal tradeoff synthetic discriminant function filter is proposed to improve the system integration and the recognition ability. The specialized digital microlens is designed for the correlator to replace the physical lens. By the theoretical analysis, the struct...
Similar publications
Signature is a biometric trait that has piqued the interest of researchers. This is due to its high rate of acceptability. Offline signature in particular, has been around for a while and hence its suitability as a biometric trait. This paper proposes an offline signature recognition system using a multiple algorithm approach. The system accepts ha...
The facial expression recognition (FER) has been an attractive research area in computer vision. An existing FER system is based on still images, facing complex problem in discriminating foreground from background cluster without motion information. To overcome this problem, the FER in video motion is projected. The execution work is based on HAAR...
Complex backgrounds, occlusions, and non-uniform classes present great challenges to pest recognition in practical applications. In this paper, we propose a multiscale attention learning network to address these problems. This network recursively locates discriminative regions and learns region-based feature representation in four branches. Three n...
Image acutance or edge contrast in an image plays a crucial role in hyperspectral hand biometrics, especially in the local feature representation phase. However, the study of acutance in this application has not received a lot of attention. Therefore, in this paper we propose that there is an optimal range of image acutance in hyperspectral hand bi...
Noise amplitude in original time domain data is usually discrete and sparse. This article presents a digital filter denoising method based on statistical frequencies of the signal values. The effective signal and noise signal are identified by comparing the frequency of the value of each pixelin the original signal with the preset validity discrimi...
Citations
... Other advantages of the VLC are the exceptionally fast time required by the entire process, the high SNR (signal-to-noise ratio) and the large values of the space-bandwidth product, together with characteristics of the matched filter such as robustness, efficiency, location accuracy and high capability of discrimination. On the other hand, this type of correlator also presents some drawbacks: the alignment of the components in the setup must be extremely precise; it is more complex than the JTC; the matched correlation filter is very sensitive both to rotations of the input image and to changes of the scale size [20,22,47,48]. ...
... The proposed filter was used in a planar integrated 2f optical correlator and was validated by experiments in a real setup. Xu et al. [48] presented an optical correlator with a 2f planar integrated structure, using an OTSDF filter, to enhance the integration of the setup and its ability to recognize patterns invariant to distortions. Instead of physical lenses, two specific digital microlenses were used (as in [7]). ...
Optical correlators are efficient optical systems that have gained a wide range of applications both in image recognition and encryption, due to their special properties that benefit from the optoelectronic setup instead of an all-electronic one. This paper presents, to the best of our knowledge, the most extensive review of optical correlators to date. The main types are overviewed, together with their most frequent applications in the newest contributions, ranging from security uses in cryptosystems, to medical and space applications, femtosecond pulse detection and various other image recognition proposals. The paper also includes a comparison between various optical correlators developed recently, highlighting their advantages and weaknesses, to gain a better perspective towards finding the best solutions in any specific domain where these devices might prove highly efficient and useful.
... 我们组一直从事光学相关识别方面的研究 [18,19] , 曾在2018年将神经网络引入光学识别, 针对范德 鲁特光学相关器应用神经网络设计了高性能光学 滤波器 [20,21] [22,23] . 图2是结合深 度神经网络和光学识别衍生出的ODNN, 其衍射 面的像素单元均是可优化的神经元, 而以光速完成 调制处理, 较好地利用了神经网络和光学识别的优 势. 在深度学习的网络中, 各层神经元之间的连接 一般是全连接或是卷积 [5] , 而在ODNN中, 各层神 经元之间的连接通过光学衍射实现, 即衍射关系决 定了映射连接 [24] . ...
ODNN (Optical diffractive neural network) uses light wave as a computing medium to perform the inference and prediction function of neural network, which has the advantages of high speed, low power consumption, and parallel processing. In this paper, an ODNN with only three layers of phase modulation is designed, and a method to improve the recognition performance of ODNN based on the first-order spectral distribution of targets is proposed. Using this method, the parameters of a three-layer ODNN are effectively optimized and the optimal pixel size, diffraction distance, and method for image preprocessing are obtained. The three-layer ODNN designed in this paper has a recognition accuracy rate of 95.3% for MNIST (handwritten digit set), compared with the 91.73% accuracy achieved by the five-layer ODNN in the reference. In addition, the system volume is greatly reduced and the system structure is simplified. Combined with the advantages of high speed and low power consumption, it has huge application potential in fields such as edge computing in the future.
... But, the relationship between the coefficients of the above empirical formulas and the structural parameters of the BLM were not built, and it took time to determine the expression coefficients by experience or simulation debugging. In our laboratory, LGP is one of the main directions for the research of micro-optical components [17][18][19][20]. In 2015, based on the small-size ILGP proposed [1], we put forward a distribution expression of the microprism on the bottom surface of the ILGP and presented the relational expressions between the coefficients of the distribution expression and the structural parameters of the ILGP [21]. ...
In our lab, the coefficient relational expressions called A ( L , H ) , B ( L , H ) , and C ( L , H ) between the coefficients in the microprism distribution expression and the structural parameters of the integrated light guide plate, which is to realize high luminance uniformity, have been proposed previously. Yet, there are some deficiencies in these coefficient relational expressions, such as numerical item redundancy and inconspicuous physical characterization. On this basis, the revised coefficient relational expressions of A ( L , H ) , B ( L , H ) , and C ( L , H ) are further studied. As a result, the numerical items of A ( L , H ) , B ( L , H ) , and C ( L , H ) are reduced by one-quarter, one-third, and one-half, respectively. The redundant high-order items are reduced, and C ( L , H ) is no longer a piecewise function. After revision, the effect of each numerical item on the coefficient values is highlighted, and it makes the physical characterization of each numerical term clearer. Moreover, by applying the revised coefficient relational expressions to acquire the microprism distribution, the luminance uniformities of all designed integrated backlight modules of 7.0 in. or less in size are totally higher than 85%. Taking a 5.0 in. integrated backlight module as an example, the performance of the integrated backlight module designed by the revised coefficient relational expressions is also improved over that before the revision. It verifies the correctness, effectiveness, and universality of the revised coefficient relational expressions, which will save a lot of design time.
... Based on the research on sub-wavelength optical design and application accumulated by our research group [10][11][12][13][14][15][16][17][18][19], this paper proposes a method based on a simple periodic binary sub-wavelength structure [12,13] to design a terahertz broadband filter. A silicon-based terahertz wideband filter with a center frequency of 4.07 THz was designed. ...
... Based on the research on sub-wavelength optical design and application accumulated by our research group [10][11][12][13][14][15][16][17][18][19], this paper proposes a method based on a simple periodic binary subwavelength structure [12,13] to design a terahertz broadband filter. A silicon-based terahertz wideband filter with a center frequency of 4.07 THz was designed. ...
In this paper, a silicon-based terahertz wideband filter was designed using rigorous couple wave theory to achieve high diffraction efficiency and wideband filtering performance. The optimal parameters, such as filter period, duty ratio, and groove depth, are given. The design value and tolerance analysis were carried out to obtain the allowable deviation range of each structural parameter of the filter. The device was fabricated and tested for performance. The experimental results were consistent with the design results, proving the feasibility of the design method and can be applied in terahertz, communication, imaging, and other fields.
... Our group has successfully applied the neural network algorithm to solve the problem of the microstructure distribution on the lower surface of the light guide plate [16], and posses many years of research experience in optical correlation recognition and micro optical design [17][18][19][20][21][22][23]. Based on the previous research, a novel method of neural network applied in optical correlation recognition is presented to enhance the poor performance in 4F optical correlator. ...
... The gray scale in the Fig. 3 is the optimized phase distribution of the filter, and its gray value 0-255 corresponds to the phase value of the filter 0-2π. The 8 layer neural network model not only reveal the physical process of the 4f optical correlator, but also is beneficial to transform the neural network for simulating the correlator with other structure in the future, such as 2f optical correlator, or oblique incident optical correlator [17], and it lays the foundation for the filter optimization of other structure correlator. ...
By introducing neural network theory into the classical 4F optical correlation recognition system and utilizing the learning principle of the neural network to optimize the optical filter, a novel method of neural network correlation recognition is presented to significantly improve the performance of the optical correlator. A new 8 layer neural network model is constructed according to the principle of the 4F optical correlator, and the phases of the optical filter are set as the connection weighting factors of the network. The new neural network is trained by distorted targets to obtain the optimal phase combination of the filter which is a pure phase device with 1080 * 1080 phase points. The novel method breaks through the limitations of the performance and design method of optical filters and overcomes the problem that the multipoint phase device is hard to optimize. Moreover, the optimized filter is applied to 4F optical correlator for the recognition of the various distortion targets. The evaluation parameters of the output correlation peak are averagely increased up to 306.7% than applying the optimal trade-off synthetic discriminant function (OTSDF) filter, and the correlation peaks are higher, sharper and easier to be distinguished, which indicates that the optical correlator with the new filter have better performance in distortion invariant recognition.
Automatic Target Recognition (ATR) using Correlation Pattern Recognition (CPR) in IoT-based applications encounters limitations like limited memory and inadequate computational resources. One reason is the required quantity of reference templates for each target/object to cover all features of a target/object. To mitigate the issue of reference templates per target/object without accuracy degradation, this paper proposes energy compaction-based CPR autoencoder-training. Additionally, a newly proposed performance metric known as Peak Energy Gain (PEG) estimates the quality of the correlation plane and the feature compression capability CPR methods. The proposed, composite filtering strategy, Eigen Maximum Average Correlation Height (EMACH), and Extended Eigen Maximum Average Correlation Height ( $E^{2}$ MACH) are vigorously validated using publicly available biometric and object image databases. By training a single reference template, the proposed training method achieves 97.97% mean accuracy with the second-best approach of $E^{2}$ MACH that attains 53.04% mean accuracy on the Pose Estimation Database. For bio-metric fingerprint verification, the mean Equal Error Rate (EER) of the proposed approach and the composite strategy is 3% and 29.69%, respectively on the FVC2002DB1A database. Similarly, the mean EER of the proposed approach and the composite strategy is 10.55% and 26.32%, respectively on the FVC2006IA database. For FEI faces dataset, the proposed method achieves 1.41% mean EER, and the composite filtering approach achieves 21.43% mean EER. On the University of Tehran Iris database, the proposed autoencoder-based methodology obtains 19.07%, and 18.07% mean EER on the left and right side iris instances, respectively. The comparative results for each dataset demonstrate superiority of AE-based method over the state-of-the-art CPR methods.
It is difficult for optical correlator to recognize the target with large rotation distortion by using existing filters. To solve this problem, a new neural network model is constructed based on the physical recognition process of planar integrated 2F optical correlator. The new optical filter is optimized by training the neural network. The planar integrated 2f optical correlator can output sharp correlation peaks by using the new filter to recognize targets with arbitrary rotation distortion. Compared with the traditional OTSDF filter, the average increase of correlation peak index is up to 1402.4%, while the recognition performance of distortion invariant is also better than the new NNCRF filter. The simulation and experimental results show that the optical filter designed in this paper can effectively solve the problem of weak recognition ability of optical correlator for rotating distortion targets, especially for large angle rotating distortion targets.
