32 results about "Gabor wavelet transform" patented technology

An innovative optical system and method is disclosed for analyzing and uniquely identifying high-order refractive indices samples in a diverse population of nearly identical samples. The system and method are particularly suitable for ultra-fine materials having similar color, shape and features which are difficult to identify through conventional chemical, physical, electrical or optical methods due to a lack of distinguishing features. The invention discloses a uniquely configured optical system which employs polarized sample light passing through a full wave compensation plate, a linear polarizer analyzer and a quarter wave retardation plate for producing vivid color bi-refringence pattern images which uniquely identify high-order refractive indices samples in a diverse population of nearly visually identical samples. The resultant patterns display very subtle differences between species which are frequently indiscernable by conventional microscopy methods. When these images are analyzed with a trainable with a statistical learning model, such as a soft-margin support vector machine with a Gaussian RBF kernel, good discrimination is obtained on a feature set extracted from Gabor wavelet transforms and color distribution angles of each image. By constraining the Gabor center frequencies to be low, the resulting system can attain classification accuracy in excess of 90% for vertically oriented images, and in excess of 80% for randomly oriented images.

The invention provides a deep learning-based multimodal emotion analysis method and system for acupuncture and moxibustion, belonging to the technical field of emotion recognition. The present invention uses Gabor wavelet transform for feature extraction on facial expression features, uses UPLBP to perform feature extraction on EEG signals, performs dimensionality reduction processing, and then uses sparse linear fusion of facial expression features and EEG signal features to form a unified , normalized eigenvectors, and transform eigenvectors into tensor form, input CNN‑LSTM network for training, remove redundant information and obtain predicted emotion classification information, and calculate network loss by comparing predicted emotion classification information with actual emotion classification information function and accuracy. The present invention fuses expression features and EEG signal features, removes redundant information, and obtains predicted emotion classification information through CNN-LSTM network training, thereby improving the accuracy of emotion recognition.

The invention provides a facial expression recognizing method based on Gabor transform optimal channel blur fusion, comprising the following steps of: S1, dividing a Gabor filter into 11 channels, carrying out Gabor wavelet transform on a facial expression image subjected to normalization by the Gabor filter to extract texture characteristics of the facial expression image; S2, establishing classifiers corresponding to the channels one by one, respectively sending the texture characteristics extracted from the channels into the classifiers for classifying and recognizing, calculating the recognition rate and the definition of each classifier to various expressions, selecting 4 classifiers with top comprehensive ranking of the recognition rate and the definition to be used as optimal classifiers; and S3, carrying blur fusion on the four optimal classifiers to obtain a facial expression recognition result. The facial expression recognition method has the advantages of small calculated amount, high calculating speed and high recognition precision.

The invention relates to a non-fluorescent eye fundus image based automatic segmentation method for retinal blood vessels. The method comprises following steps: step one, an image is preprocessed, blood vessel characteristics are enhanced, and background noise is weakened; step two, morphological characteristics of the retinal image are extracted with a multi-scale linear operator, and a refined blood vessel framework is segmented with two-dimensional Gabor wavelet transform; step three, after the refined blood vessel framework is segmented, the image is subjected to path morphology filtering, the maximum path length is combined with geometrical characteristics of a blood vessel region, and an adjacent map is constructed; step four, a binaryzation process is performed with regional connectivity analysis and a hysteresis threshold technology for auxiliary segmentation of microvessels, and the segmentation is finished. Compared with the prior art, the method has the advantages as follows: the blood vessel location accuracy is higher, the adhesion phenomena are reduced, and a retinal blood vessel network of the eye fundus image is effectively segmented.

The invention discloses a multi-temporal high-resolution remote sensing image building extraction method based on a multi-feature LSTM network, which belongs to the technical field of satellite remote sensing image processing and application. The purpose is to solve the problems of low accuracy rate, high misclassification rate and blurred boundary of building extraction results in existing methods. The present invention uses multiple multi-temporal Gaofen No. 2 remote sensing images as the data source, uses the method based on HSI color transformation to extract the spectral features of the building, and extracts the shape features of the building based on the method of combining image segmentation and conditional random field post-processing. The method based on Gabor wavelet transform to extract the texture information features of the building and the method based on the DSBI index to extract the index features of the building, and the extracted spectrum, shape, texture and index features of the multi-temporal buildings are composed of 60 feature bands The building feature set, and the produced building samples and labels are sent to the LSTM network to obtain the rough extraction results of the buildings, and the final results are obtained after morphological processing.

The invention provides a method and a system for identifying human facial expression. The method for identifying the human facial expression comprises the steps of utilizing ASM (active shape model) algorithm to extract a shape model and feature points of a to-be-identified facial image, then, utilizing Gabor wavelet transform algorithm to extract amplitude texture characteristics of the feature points; obtaining a to-be-identified eigen face through projection for the amplitude texture characteristics; and identifying expression of a to-be-identified human face by comparing similarity between the to-be-identified eigen face and a sample eigen face. According to the method and the system for identifying human facial expression, the ASM algorithm is combined for extracting the feature points, so that identification through the Gabor wavelet transform algorithm can be performed only by comparing textural characteristics of the feature points. Compared with the existing facial identification technology only adopting the Gabor wavelet algorithm, the method and the system for identifying the human facial expression has the advantages of small data processing capacity, short identification time, low internal memory loss and the like.

The present invention provides an En-ULLELDA-based (ensemble-unified locally linear embedding and linear discriminant analysis) method of multi-view model recognition. The method comprises the following steps: firstly, selecting a vehicle ROI (region of interest); secondly, extracting the high-level information that can reflect the model as features through Gabor wavelet transform; then, carrying out the dimension reduction of features on the basis of a novel manifold learning algorithm En-ULLELDA, so as to transform the features to a low-dimensional space; and obtaining the corresponding models from the features of a test sample by a classifier during the recognition. According to test results, the invention can effectively handle the problem of multi-view model recognition on urban roads; and the invention can eliminate the effects of parameter variation on the recognition results by employing the integration technology, so that the invention has higher robustness.

The invention utilizes Gabor wavelet decomposition to extract the texture features of the interphalangeal joints of four fingers (index finger, middle finger, ring finger and little finger). Feature extraction will eventually result in a covariance matrix between Gabor wavelet subbands. The covariance matrix can describe the correlation structure between the Gabor wavelet subbands, and it has better discrimination than calculating the subband variance in an independent way; the identification part is to compare the current acquisition identity Gabor subband covariance matrix with the memory Comparison between Gabor subband covariance matrices of registered identities. In order to improve the performance, the Gabor wavelet sub-band is firstly transformed by CDF (cumulative distribution function), and then the transformed result is calculated. Main features of the present invention: (1) The identification method has high security, and the characteristics of the four fingerprints ensure the security of the system; (2) It has good resistance to noise, and Gabor wavelet transform and CDF projection can effectively resist image noise interference .

A method and system for forming a facial expression intensity calculation model. The model is used to estimate the facial expression intensity under expression classification. First, an expression database is obtained, image data in the expression database are preprocessed, and facial part data is extracted. The feature extraction of the three modes of facial geometric features, local binary mode and Gabor wavelet transform is carried out respectively, and the data output in the previous step are trained by using the fully supervised mode, semi-supervised mode and unsupervised mode respectively, and the features and facial expression strength are obtained The data formed after training are respectively used as the input of the ordinal random forest algorithm for training, and the facial expression intensity calculation sub-models are respectively obtained, and the final facial expression intensity calculation model is formed according to each sub-model. The invention utilizes the database to train the facial expression intensity calculation model under the expression classification, and uses the model to process the image data to obtain the intensity under the trained expression classification.