Настройки

Укажите год
-

Небесная энциклопедия

Космические корабли и станции, автоматические КА и методы их проектирования, бортовые комплексы управления, системы и средства жизнеобеспечения, особенности технологии производства ракетно-космических систем

Подробнее
-

Мониторинг СМИ

Мониторинг СМИ и социальных сетей. Сканирование интернета, новостных сайтов, специализированных контентных площадок на базе мессенджеров. Гибкие настройки фильтров и первоначальных источников.

Подробнее

Форма поиска

Поддерживает ввод нескольких поисковых фраз (по одной на строку). При поиске обеспечивает поддержку морфологии русского и английского языка
Ведите корректный номера.
Ведите корректный номера.
Ведите корректный номера.
Ведите корректный номера.
Укажите год
Укажите год
Применить Всего найдено 8. Отображено 8.
06-08-2014 дата публикации

Large-scale facial expression recognition method based on multiscale LBP and sparse coding

Номер: CN103971095A
Автор: PENG XIANLIN, XIA ZHAOQIANG, FENG XIAOYI, PENG JINYE, WANG JUN, MAO XIAOFEI, CUI MINGHUI, HU XUTAO
Принадлежит:

The invention provides a large-scale facial expression recognition method based on a multiscale LBP and sparse coding. The method comprises the steps that a large-scale facial expression database is built firstly, a training database and a testing database are generated based on a random sampling technology, then facial expression features are expressed through multiscale LBP features, then a dictionary needed in a sparse coding method is generated, a new expression sample is solved to obtain an optimal sparse coefficient, and the sparse coefficients of different expressions are accumulated to recognize the expression samples. According to the method, a high-robustness feature expressing mode is obtained through the multiscale LBP features, the sparse problem in large-scale facial expression recognition is solved through sparse coding, and the effectiveness of the large-scale facial expression recognition method based on the multiscale LBP and sparse coding is verified.

Подробнее
Номер записи: 1
15-08-2023 дата публикации

Unsupervised bidirectional variational self-encoding essential image decomposition network, method and application

Номер: CN116597274A
Автор: JIANG XIAOYUE, HAN YIFEI, XIA ZHAOQIANG, FENG XIAOYI
Принадлежит:

The invention relates to an unsupervised bidirectional variational self-encoding essential image decomposition network, method and application, and belongs to the technical field of image recognition. The network uses a parallel encoding and decoding strategy, that is, two sets of same but unrelated deep learning networks consisting of encoders and decoders are used for generating images of reflection maps or illumination maps. The encoder and the decoder are composed of a network module containing down-sampling or up-sampling; between the encoder and the decoder, the features pass through the encoder to obtain a mean value and a standard deviation of potential vector distribution, and then the features conforming to specific distribution are sampled from the mean value and the standard deviation. According to the method, the idea of introducing variational self-encoding is put forward, an intermediate vector in the encoding and decoding process is constrained, and the robustness of a model ...

Подробнее
Номер записи: 2
15-08-2023 дата публикации

Self-attention-based multi-scale coding and decoding essential image decomposition network, method and application

Номер: CN116597273A
Автор: JIANG XIAOYUE, HAN YIFEI, XIA ZHAOQIANG, FENG XIAOYI
Принадлежит:

The invention relates to a self-attention multi-scale coding and decoding essential image decomposition network and method and application, and belongs to the technical field of image recognition. The network uses a parallel encoding and decoding strategy, namely, two sets of encoders and decoders which are the same but are not associated with each other are used for generating images of a reflection map and an illumination map. Each network comprises an encoder, a decoder and a hop link. Wherein each of the encoder and the decoder consists of a plurality of self-attention modules and a down-sampling or up-sampling module. The self-adaptive window self-attention provided by the method can adaptively adjust the window according to the image content so as to enhance the attention of the network to different types of features and reduce the resource occupation of the model; and the channel enhancement self-attention module is used for enhancing effective information in the feature map by expanding ...

Подробнее
Номер записи: 3
08-10-2014 дата публикации

Image quality evaluation method based on sparse structure

Номер: CN104091343A
Автор: WU JUN, LI HUIFANG, FENG XIAOYI, XIA ZHAOQIANG, CAO ZHENGWEN
Принадлежит:

The invention discloses an image quality evaluation method based on a sparse structure. The method is used for solving the technical problem that the evaluation effect of an existing image quality evaluation method is poor. According to the technical scheme, firstly, input reference images and input degraded images are sampled to obtain a reference image sampling matrix and a degraded image sampling matrix; then, a dictionary is obtained in a studying mode through the reference image sampling matrix, and in the process of working out a sparse solution, sparse representation is carried out on the reference image sampling matrix and the degraded image sampling matrix through the dictionary obtained in the studying mode to obtain a reference image sparse representation coefficient matrix and a degraded image sparse representation coefficient matrix; finally, the image quality is evaluated according to the change degree of the sparse coefficient structure. According to the method, by the adoption ...

Подробнее
Номер записи: 4
06-06-2023 дата публикации

Attack image defense method based on denoising and super-resolution reconstruction fusion

Номер: CN116228537A
Автор: JIANG XIAOYUE, WANG ZHONGPENG, FENG XIAOYI, XIA ZHAOQIANG, HAN YIFEI
Принадлежит:

The invention relates to an attack image defense method based on denoising and super-resolution reconstruction fusion, and the method comprises the steps: firstly carrying out the coarse cleaning of noise in an image through employing a multi-stage nested coding and decoding network; the noise is mainly high-frequency information, so that the multi-scale coding and decoding network can guide the reconstruction of the high-frequency information by utilizing the low-frequency information, and the elimination of the high-frequency noise is further realized. And then, fine cleaning is performed on the noise by using a super-resolution reconstruction network, and distribution of residual adversarial noise is destroyed by injecting a high-frequency component in a super-resolution reconstruction process. According to the method, self-encoder denoising and super-resolution reconstruction are effectively combined to carry out confrontation and defense. In an experiment, the sample cleaning defense ...

Подробнее
Номер записи: 5
04-07-2023 дата публикации

Reprogramming resisting model and method

Номер: CN116383820A
Автор: FENG XIAOYI, ZHENG YANG, JIANG XIAOYUE, XIA ZHAOQIANG, DANG SIHANG, WU JIAPING, DANG CHEN, ZHANG FENG, GAO JIE, KANG JIAN
Принадлежит:

The invention belongs to the field of computer vision and computer security, and particularly relates to an anti-reprogramming model and method, the anti-reprogramming model comprises a feature extraction module, a classifier, an anti-disturbance function, an anti-reprogramming linear function and an anti-loss function; the classifier comprises a plurality of convolutional neural network models; the feature extraction module extracts an initial feature map in the target domain data set image by adopting a convolutional neural network model, and obtains adversarial data through an adversarial disturbance function based on the initial feature map; the adversarial reprogramming linear function is designed based on adversarial data; the anti-reprogramming linear function enables the target domain data set images to be classified in the target detection data set according to the intention of an attacker; according to the anti-reprogramming method, by evaluating the classification result of the ...

Подробнее
Номер записи: 6
06-08-2014 дата публикации

Area-of-interest detection method based on Kinect

Номер: CN103971116A
Автор: PENG XIANLIN, XIA ZHAOQIANG, FENG XIAOYI, PENG JINYE, WANG JUN, MAO XIAOFEI, CUI MINGHUI, HU XUTAO
Принадлежит:

The invention discloses an area-of-interest detection method based on Kinect. The area-of-interest detection method is used for solving the technical problem that an existing area-of-interest detection method based on an improved saliency map model is low in accuracy. According to the technical scheme, a Kinect3D camera lens is used for obtaining a two-dimensional RGB image and depth information; on the basis, the RGB image is used for extracting various visual features and establishing a multi-scale visual feature map; then the feature map and a depth map are fused to generate the saliency map, and a winner-take-all strategy is used for generating a two-value saliency map; finally, dilation is performed on the two-value saliency map to detect a final area of interest. The 3D image, generated by the Kinect camera lens, in a RGB-D format can be used for detecting the area of interest, wherein the area of interest is consistent with the result of perception through the human eyes. Under the ...

Подробнее
Номер записи: 7
30-05-2023 дата публикации

Essential image decomposition method based on bilateral feature pyramid network and multi-scale identification

Номер: CN116188791A
Автор: JIANG XIAOYUE, WANG ZHONGPENG, FENG XIAOYI, XIA ZHAOQIANG, HAN YIFEI
Принадлежит:

The invention relates to an essential image decomposition method based on a bilateral feature pyramid network and multi-scale identification, and the method comprises the steps: taking two parallel generative adversarial networks as a main network, and carrying out the reconstruction of a reflection image and an illumination image; for the generative network, a strategy of local partial frequency feature fusion is provided, and selection and reservation of high-frequency reflection features and low-frequency illumination features are realized respectively. Meanwhile, a multi-scale adaptive combination module is added in the discriminator, adaptive evaluation is carried out on contribution of multi-scale features, the discrimination effect is enhanced, and the generation effect is improved. The method disclosed by the invention is excellent in performance on various data sets, and in the MPI-Single data set, compared with other methods, the reconstruction mean square error of the optimal ...

Подробнее
Номер записи: 8