120 results about "Visual cortex" patented technology

An artificial synapse array and virtual neural space are disclosed.More specifically, a cognitive sensor system and method are disclosed comprising a massively parallel convolution processor capable of, for instance, situationally dependent identification of salient features in a scene of interest by emulating the cortical hierarchy found in the human retina and visual cortex.

Reading disorders are diagnosed and remediated in a subject by respectively measuring and improving contrast sensitivity for motion discrimination of the subject. A background is displayed on a monitor with a contrast and a spatial frequency. A test window is superimposed over the background and includes a test pattern with a contrast and a spatial frequency. The contrasts and the spatial frequencies are within respective ranges which stimulate the visual cortical movement system of the subject. The test pattern is then moved within the test window. The subject provides a signal indicative of the direction the subject believes the test pattern moved. In response to this signal, the contrast of the test pattern, the spatial frequency of the background, or the spatial frequency of the test pattern is modified, either by increasing or decreasing its respective value. This process is then repeated a number of times, cycling through predetermined combinations of test patterns and backgrounds. Contrast sensitivity may be measured to determine whether a child is dyslexic. Repeated stimulation by the methods and apparatus of the invention improves contrast sensitivity, thereby remediating dyslexia and improving reading ability.

A method, device and system for stimulating visual tissue, typically in the retina or visual cortex, to achieve an artificial percept of light or image. The method includes providing stimulating electrodes suitable for placement in proximity to the visual tissue and generating a series of short-duration stimulation signals having a duration of less than about 0.5 milliseconds each. The short-duration stimulation signals are applied through the stimulating electrodes with varying frequencies that are substantially matched to a spiking range of frequencies of at least one ganglion cell for perceiving brightness or image.

The present invention is a method of improving the persistence of electrical neural stimulation, and specifically a method of improving the persistence of an image supplied to a retina, or visual cortex, through a visual prosthesis. A continuously stimulated retina, or other neural tissue, will desensitize after a time period in the range of 20 to 150 seconds. However, an interruption of the stimulation on the order of a few milliseconds will restore the retinal sensitivity without the user perceiving the interruption, or with the user barely perceiving the interruption.

The invention discloses a method for extracting a face feature by simulating a biological vision mechanism, and belongs to the field of image processing and pattern recognition. The method comprises the following steps of: simulating a learning mechanism of primary visual cortex simple cells, training a group of filters for describing a simple cell receptive field, and selecting a small number ofillumination invariant features of image extracted by the filter, which have the specific frequency selectivity, from the group of filters; simulating the function of primary visual cortex complex cells, and adding the expression and the translational invariance of the feature on the basis of the illumination invariance; strengthening a salient region of the invariant feature by using a visual attention mechanism; and converting the strengthened invariant feature into a characteristic vector for face recognition. An experiment shows that the method can effectively reduce the influence of illumination, expression and translation change on face recognition effect, and has real-time processing capability.

The invention discloses an image processing device of visual cortex prosthesis. The image processing device comprises a left miniature camera, a right miniature camera, an image preprocessor and an image coder, wherein the left and right miniature cameras output respectively and synchronously collected video information in real time to the image preprocessor; the image preprocessor splices and merges two paths of received video information into one full-view image, and further completes the preprocessing operation such as image filtering, enhancing and pixel reduction, and outputs the image to the image coder; the image coder performs further processing and coding on the preprocessed full-view image and converts the image into an electric signal capable of being received by the human body visual cortex. The image processing device provided by the invention realizes the goal of converting the image information collected by double cameras into a current signal used for irritating the visual cortex; on the premise of ensuring the nerve cells to issue the equivalent action potential, the intensity of the incitation current can be effectively reduced, so that the goal of effectively reducing the brain tissue injury can be achieved.

An artificial visual recognition system and method employ a digital processor and a model executed by the digital processor. The model has a loose hierarchy of layers. Each layer, from a lowest hierarchy level to a top level, provides relatively increasing selectivity and invariance of the input image. The hierarchy allows bypass routes between layers. On output, the model produces feature recognition and classification of an object in the input image. In some embodiments, windowing means provide windows of the input image to the model, and the model responds to shape-based objects in the input image. In another feature, segmenting means segment the input image and enables the model to determine texture-based objects in the input image.

A method for creating sensory experiences operates by scanning the acoustical signal across the human neural cortex to create the desired sensory perceptions. The acoustic signal is scanned in a predetermined pattern. The pattern is then modified to fill in spaces in the predetermined pattern so that over a short time period, the desired signal is scanned across the intended region of the neural cortex. In one exemplary embodiment, the pattern begins with an array of points on the cortex. Thus, an acoustic signal in an array of points is directed towards the cortex. The acoustic pattern is then shaped to expand in radius about each point. Thus, the acoustic signal scans the visual cortex in an array of expanding circles. Varying the signal at each point along the radius as it expands produces neural firing differences in the neural tissue. When the circles expand to where they begin to touch, the pattern changes to fill in the areas between the original array of points. The new circles are centered about the points between the original stimulation locations, and the acoustic signal contracts about these new centers. The signal continues to contract about the new center points. When the new circles have contracted to an array of points, the process can be repeated from the start or simply reversed. Another method operates by forming concentric circles and expanding and contracting each of the concentric circles to fill in the original spaces between the concentric circles.

The invention belongs to the field of high resolution remote sensing image processing and particularly relates to a high resolution remote sensing image local feature extraction method based on 2D-Gabor. According to the method provided by the invention, a scale space pyramid expression of an image is firstly established; accelerated partition testing features of different feature scales are searched in the scale space, and a maximum value inhibition method is utilized to obtain a feature point and to determine the position and the scale of the feature point; then a local feature descriptor based on a binary system is established; and finally, a Hamming distance is used in a similarity measurement method to perform feature matching of images of the same scene under different perspective conditions, then an RANSAC algorithm is adopted to perform feature purification, and error matching point pairs are removed. The method provided by the invention can accurately simulate cognitive features of the visual cortex and the retina of human beings. In the feature detection process, an invariance property for change in brightness and scale is achieved, and optimal performances can be obtained at the same time in the time domain and the frequency domain.

Disclosed herein is the treatment of vision loss in a mammal by transplanting an effective amount of hNT-Neuron cells. The treatment can be accomplished by injecting the cells into the retinal area of the eye. Additionally, the cells can be injected into the visual cortex of the brain. Conditions to be treated are vision loss due to optic nerve damage, including glaucoma, optic nerve sheath meningioma and glioma, Graves' ophthalmopathy, benign or malignant orbital tumors, metastatic lesions, tumors arising from the adjacent paranasal sinuses or middle cranial fossa, giant pituitary adenomas, brain tumors or abscesses, cerebral trauma or hemorrhage, meningitis, arachnoidal adhesions, pseudotumor cerebri, cavernous sinus thrombosis, dural sinus thrombosis, encephalitis, space-occupying brain lesions, severe hypertensive disease or pulmonary emphysema.

The present invention discloses a no-reference image quality evaluation method based on structure similarity mapping dictionary learning. The method comprises the steps of: performing joint training of a structure similarity mapping image in a dictionary training image set and a reference image to obtain a joint dictionary, employing the joint dictionary to perform sparse decomposition of distorted images in an evaluation training image set and an evaluation test image set to obtain a structure similarity mapping graph, extracting gray co-occurrence matrixes in various directions from the structure similarity mapping graph, and converting the gray co-occurrence matrixes to vectors, calculating the standard deviation, the skewness and the kurtosis of the vectors of the gray co-occurrence matrixes in multiple scales to combine feature vectors, sending the feature vectors to a support vector regression for training and test, and performing prediction to obtain an objective quality evaluation score. The method performs training to obtain the structure similarity mapping dictionary, employs the structure similarity mapping dictionary for feature extraction and is identical to nerve characteristics of a brain visual cortex so as to obtain a no-reference image quality evaluation result with a more accurate prediction effect.

A perimeter for examining a visual field condition of an examinee's eye, includes: a target presenting unit which presents a stimulation target for examination to the eye, a presentation condition of the target including a presentation position of the target being variable; a response unit with which the examinee responds when the examinee can perceive the presented target; brain activity detecting unit which detects an activity condition of a visual cortex of the examinee; and an operation unit which determines if the response with the response unit with respect to presentation of the target is resulted from a fact that the examinee can perceive the target based on a detection result of the brain activity detecting unit and which obtains the visual field condition on the basis of the presentation condition of the target and the response result in consideration of the determination result of the brain activity detecting unit.

An information acquiring method to acquire brain activity information from biological subjects presented a sensory stimulus, including the following processes,(1) a process to present the sensory stimulus to at least one brain area of the biological subjects including the visual cortices of the cerebral cortex, the auditory cortices of the cerebral cortex, and the vestibular cortices of the cerebral cortex,(2) a process to measure brain activity in multiple, different regions of the brain areas while the sensory stimulus is presented, and(3) a process to evaluate stress provided to the subject by the sensory stimulus based on the brain activity data for the multiple regions.

The invention relates to an image quality evaluation method based on independent component analysis. The image quality evaluation method is characterized by being suitable for image quality evaluation of a grey scale map and a color image synchronously. The image quality evaluation method comprises the following steps of firstly, centrally training a group of ICA (Independent Component Analysis) decomposing matrixes from a reference image by utilizing a FastICA (Fast Independent Component Analysis) algorithm; secondly, multiplying each image block in the reference image and an image to be evaluated, and the ICA decomposing matrixes so as to obtain the independent component of each image block; lastly, measuring the quality of the image to be evaluated according to the difference of the independent components of the reference image and the image to be evaluated. In comparison with the conventional method, the method is capable of simulating expression of a visual signal in a human visual cortex and is closer to subjective image quality evaluation. The main calculated quantity of the method is centralized to the independent components, which are obtained by multiplying each split image block and the ICA decomposing matrixes, of the image blocks, but the calculation of each image block is independent, so that parallel computing is adopted, thus the execution efficiency is improved.

An object recognition system and method is provided which uses automated algorithmically determined negative training Negative training with respect to a particular object classifier allows for more streamlined and efficient targeted negative training, enabling time and cost savings while simultaneously improving the accuracy of recognition based on the targeted negative training According to certain additional aspects, embodiments of the invention relate to a system and method for structuring an object recognition engine to maximize recognition accuracy and emulate human visual cortex.

The invention discloses an aurora image classification method based on biological stimulation characteristics and manifold learning. The method comprises the following steps of: (1) carrying out preprocessing of edge denoising on an input aurora image; (2) carrying out Gabor filtering on the aurora image subjected to preprocessing by using a multi-directional Gabor filter group, so as to obtain C1-layer characteristic graphs, and taking the sum of pixel gray level values of each characteristic graph as a C1 characteristic of the aurora image; (3) extracting a Gist characteristic of the aurora image; (4) fusing the C1 characteristic and the Gist characteristic so as to obtain a BIFs characteristic of the aurora image; (5) carrying out fuzzy C-mean value clustering on the BIFs characteristic, and subsequently carrying out dimensionality reduction by using a manifold learning algorithm so as to obtain the expression of the BIFs in a low-dimension space; and (6) respectively classifying aurora images by using a support vector machine (SVM) and a nearest neighbor (NN) classifier. By utilizing the method, the recognition process of human brain visual cortex can be well simulated, data redundancy is reduced, classification accuracy rate is improved, and therefore the method can be used for scene classification and object recognition.

An information acquiring method configured to acquire brain activity information from biological subjects presented a sensory stimulus, in which the brain activity information is acquired by measuring the brain activity in at least one region of the brain areas of the biological subjects including the visual cortices of the cerebral cortex, the auditory cortices of the cerebral cortex, and the vestibular cortices of the cerebral cortex; while the stimulus which changes over time is presented to the brain areas by the stimulus presenting unit before and after the sensory stimulus is presented to the brain areas.

The invention discloses an SAR image man-made target detection method based on visual contrast and information entropy. Local contrast information of an image is extracted by simulating simple cells in a primary visual cortex according to biological visual features; a computing pool model is learnt and S units are integrated into a complex unit C, and Gaussian filtering processing is performed so that a global salient graph is generated; threshold segmentation is performed on the salient graph according to the maximum information entropy principle; isolated pixel sets are eliminated by morphological filtering processing, and connectivity of the pixel sets is enhanced; and target edge information is extracted and displayed on an original SAR image so that a target detection result is obtained. The achieved beneficial effects are that effective detection of the man-made target in the SAR image can be realized, computation burden is low, processing efficiency is high and project implementation is facilitated, and thus the method can be used for rapid detection of the small man-made target.

The invention relates to a microelectrode array stimulation device that can stimulate visual cortical nervous tissue and the stimulation method. The device can output electric stimulation signal alternatively through different stimulation points under action of control circuit, to stimulate visual cortical nervous tissue. The device comprises microelectrode array chip, signal path control part, stimulation signal generation part and coding part. The microelectrode array chip is constructed with a plurality of stimulation unit arrays on one base plate, each stimulation unit is corresponding to one stimulation position on visual cortical nervous tissue, and includes a plurality of ultra micro-stimulation points formed with a plurality of microelectrodes. The signal path control part selects the connection way of stimulation signal and ultra-micro stimulation point in stimulation unit and control visual cortical stimulation mode, couples stimulation signal to visual cortical nervous tissue through selected ultra-micro electrode couple. The invention can realize automatic stimulation for visual cortical nervous tissue and increase life length for stimulation electrode.

An artificial visual recognition system and method employ a digital processor and a model executed by the digital processor. The model has a loose hierarchy of layers. Each layer, from a lowest hierarchy level to a top level, provides relatively increasing selectivity and invariance of the input image. The hierarchy allows bypass routes between layers. On output, the model produces feature recognition and classification of an object in the input image. In some embodiments, windowing means provide windows of the input image to the model, and the model responds to shape-based objects in the input image. In another feature, segmenting means segment the input image and enables the model to determine texture-based objects in the input image.

The invention discloses an image classification method based on a visual cortex processing mechanism and pulse supervised learning. The image classification method based on a visual cortex processingmechanism and pulse supervised learning includes the steps: 1) inputting an image in a mode of dynamic image, and then processing the input under three different space-time scales; 2) solving responseof a V1 layer simple type Neuron; 3) solving response of a V1 layer complex type Neuron; 4) solving response of a V4 layer Neuron; 5) training output layer connection; and 6) inputting a test sample.The image classification process of the image classification method based on a visual cortex processing mechanism and pulse supervised learning is closer to the processing process of a real brain, and the image classification method can preferably extract the local orientation information of the input image so as to preferably perform image classification. Besides, the image classification only needs training connection of the output layer without training layer by layer, thus having higher efficiency. For the image classification method, classification accuracy on a hand-written digital setis around 96%, and the classification accuracy is high.

The invention discloses an electroencephalogram (EEG) signal virtual-keyboard design method based on multiple modes. An event-related potential (ERP) of the method is a special brain-evoked potential,and reflects nerve-electrophysiology changes of a brain in a cognitive process. SSVEP is periodic response of the brain to external visual stimuli modulated at certain frequency (greater than 6Hz), and EEG of a brain visual-cortex under the paradigm shows distinct peak value features at the stimulation frequency and harmonics thereof. Early components of N1 and P2 of ERP of all brain regions aresignificantly different from amplitudes in an incubation period, and also have differences in scalp distribution. SSVEP can extract frequency spectrum features of 12Hz, 15Hz, 18Hz and 20Hz through a canonical correlation analysis (CCA) method, thus the early components of N1 and P2 of ERP can be used as row coordinates of a virtual keyboard, and the frequency spectrum features corresponding to SSVEP can be used as column coordinates of the virtual keyboard, and accuracy of the virtual keyboard based on EEG signals is greatly increased.

The invention discloses a target detection method based on a sparse representation and visual cortex attention mechanism. The target detection method is realized by the following steps of: (1) utilizing a multi-scale and multi-directional Gabor filter group to carry out Gabor filtering on an input image to be detected and extracting to obtain S1 layer cells of an HMAX model; (2) utilizing independent component analysis (ICA) and utilizing C1 layer cells of the HMAX model to construct a sparse group; and (3) utilizing C2 layer cells which have invariance on scales and translation, and a Shannon entropy theory to obtain information entropy of the C2 layer cells, and furthermore, constructing a measure for detecting a visual significance region to obtain an obvious image of a previous image. According to the target detection method disclosed by the invention, the perception property of human eyes in finding the significance of a natural scene image can be better satisfied, so that the accurate rate is higher and the false detection rate is lower; and compared with the traditional image detection method, the target detection method has the advantages that the effectiveness and the accuracy in a psychological stimulation mode image are better.

The invention discloses a method of estimating scene light source colors of a color image. The method of estimating the scene light source colors of the color image comprises extracting color components in the image, calculating a component of single-antagonism color space, calculating a response value of a double-antagonism color space, transforming the double-antagonism color space into three primary color space and calculating light source colors in the three primary color space. Because the size and the amplitude value of a difference of Gaussian (DoG) filter are flexibly selected in the double-antagonism color space so as to simulate color sensitivity of a primary visual cortex and receptive field characteristics of a double-antagonism neuron, the light source color value of an image scene can be fast and efficiently calculated according to the method, and the scene light source color value finally acquired can be used for a plurality of follow-up computer visual applications such as tone correcting and white balance. The method of estimating the scene light source colors of the color image has the advantages of fast and efficiently working out information of the light source colors of the image of each scene through simple and flexible parameter selection.

The method is applicable to the technical field of deep learning, and provides a deep learning model training method, device and equipment and a storage medium. The method comprises the following steps: according to the received training image set, using a preset feature extraction algorithm and a pre-constructed untrained deep learning model to extractcorresponding target features, performing local training on a hidden layer of the deep learning model through a target feature set formed by the extracted target features; after local training is completed, selecting a target image set from thetraining image set; performing classification training on the full connection layer of the deep learning model. The deep learning model is trained through the deep learning model, so that the number of samples for training the deep learning model is reduced, the trained deep learning model better conforms to the human brain visual cortex characteristics, the noise resistance and the displacement resistance of the deep learning model are improved, and the training speed and the training effect of the deep learning model are improved.

The invention relates to a motor imagery electroencephalogram feature enhancement method and system. The enhancement method comprises the following steps: acquiring a multi-mode neural feedback training motor imagery electroencephalogram feature enhancement paradigm; based on the multi-mode neural feedback training motor imagery electroencephalogram feature enhancement normal form, acquiring multi-channel electroencephalogram signals of a pair of a parietal region motor cortex and an occipital region visual cortex; preprocessing the electroencephalogram signals, and determining the preprocessed electroencephalogram signals; conducting dynamic and coupled multi-level feature extraction on the preprocessed electroencephalogram signals, and determining dynamic features and coupling features of the preprocessed electroencephalogram signals; and according to the dynamic features and the coupling features, displaying feature values of the motor imagery electroencephalogram signals before enhancement and feature values of the motor imagery electroencephalogram signals after enhancement. By adopting the enhancement method and system provided by the invention, the recognition rate of effective scalp electroencephalogram signals can be improved.

The invention discloses a stereoscopic vision imaging device and stereoscopic vision stimulation equipment. The device comprises a mirror used for converting the transmission direction of the light beams of a left-eye image and the light beams of a right-eye image from a horizontal direction to a vertical direction, a left-eye light path channel extending along the vertical direction and used for receiving and transmitting the light beams of the left-eye image reflected by the mirror and carrying out imaging on the exit surface of the left-eye light path channel, a right-eye light path channel extending along the vertical direction and used for receiving and transmitting the light beams of the right-eye image reflected by the mirror and carrying out imaging on the exit surface of the right-eye light path channel, and a support device on which the left-eye light path channel and the right-eye light path channel are arranged movably along the left-right direction and the vertical direction and the mirror is also arranged. The device helps to provide real stereoscopic vision stimulation close to that accepted by the human eye, and provides a novel visual basic research means for related research and test of the visual cortex area in a high-magnetic-field environment.