1973 results about "Deconvolution" patented technology

In the systems and methods of the present invention a multifocal multiphoton imaging system has a signal to noise ratio (SNR) that is reduced by over an order of magnitude at imaging depth equal to twice the mean free path scattering length of the specimen. An MMM system based on an area detector such as a multianode photomultiplier tube (MAPMT) that is optimized for high-speed tissue imaging. The specimen is raster-scanned with an array of excitation light beams. The emission photons from the array of excitation foci are collected simultaneously by a MAPMT and the signals from each anode are detected using high sensitivity, low noise single photon counting circuits. An image is formed by the temporal encoding of the integrated signal with a raster scanning pattern. A deconvolution procedure taking account of the spatial distribution and the raster temporal encoding of collected photons can be used to improve decay coefficient. We demonstrate MAPMT-based MMM can provide significantly better contrast than CCD-based existing systems.

The invention discloses an RGB-D image object detection and semantic segmentation method based on a deep convolution network, which belongs to the field of depth learning and machine vision. According to the method provided by the technical scheme of the invention, Faster-RCNN is used to replace the original slow RCNN; Faster-RCNN uses GPU, which is fast in the aspect of feature extracting, and at the same time generates a regional scheme in the network; the whole training process is training from end to end; FCN is used to carry out RGB-D image semantic segmentation; FCN uses a GPU and the deep convolution network to rapidly extract the deep features of an image; deconvolution is used to fuse deep features and shallow features of the image convolution; and the local semantic information of the image is integrated into the global semantic information.

In embodiments of the invention, an apparatus may include a display comprising a plurality of pixels. The apparatus may further include a computer system coupled with the display and operable to instruct the display to display a deconvolved image corresponding to a target image, wherein when the display displays the deconvolved image while located within a near-eye range of an observer, the target image may be perceived in focus by the observer.

A source deconvolutional network is adaptively trained to perform semantic segmentation. Image data is then input to the source deconvolutional network and outputs of the S-Net are measured. The same image data and the measured outputs of the source deconvolutional network are then used to train a target deconvolutional network. The target deconvolutional network is defined by a substantially fewer numerical parameters than the source deconvolutional network.

The invention discloses a multi-scale image semantic segmentation method. The method comprises the following steps: obtaining a to-be-segmented image and a corresponding label; constructing a full convolutional deep neural network, wherein the full convolutional deep neural network comprises a convolution module, a hole convolution module, a pyramid pooling module, a 1 * 1 * depth convolution layer and a deconvolution structure; setting hole convolution as channel-by-channel operation, and utilizing low-scale, medium-scale and high-scale characteristics in a targeted mode; training the full convolutional deep neural network, establishing a loss function, and determining parameters of the full convolutional deep neural network by training the sample image; and inputting the to-be-segmentedimage into the trained full convolutional deep neural network to obtain a semantic segmentation result. By means of the method, the image semantic segmentation problem with complex details, holes andlarge targets can be well solved while the calculated amount and the parameter number are reduced, and the consistency of category labels can be reserved while the target edges can be well segmented.

The invention discloses a single-image super-resolution reconstruction method based on a symmetric depth network and belongs to the image processing technology field. The method mainly comprises the following steps of 1, making a high resolution image block and low resolution image block training set; 2, constructing a symmetric convolution-deconvolution depth network used for model training; 3, based on the constructed depth network and the made data set, carrying out network model training; and 4, based on a learned model parameter, inputting one low resolution image, wherein acquired output is a reconstructed high resolution image. In the invention, a convolution layer and a deconvolution layer are combined and simultaneously a network depth is increased; the network depth is used to increase network performance; a reconstruction capability of an image detail portion is enhanced and a good image super-resolution reconstruction effect is acquired. The method has a wide application prospect in fields of image high definition displaying, medical imaging, a remote sensing image and the like.

An extended depth of field is achieved by a computational imaging system that combines a multifocal imaging subsystem for producing a purposefully blurred intermediate image with a digital processing subsystem for producing a recovered image having an extended depth of field. The multifocal imaging system preferably exhibits spherical aberration as the dominant feature of the purposeful blur. A central obscuration of the multifocal imaging subsystem renders point-spread functions of object points more uniform over a range of object distances. An iterative digital deconvolution algorithm for converting the intermediate image into the recovered image contains a metric parameter that speeds convergence, avoids stagnations, and enhances image quality.

A method and system is provided for enhanced ultrasonic detection and imaging of small defects inside or at the surface of an object. The Synthetic Aperture Focusing Technique (SAFT) has been used to improve the detectability and to enhance images in conventional ultrasonics and this method has recently been adapted to laser-ultrasonics. In the present invention, an improved version of the frequency-domain SAFT (F-SAFT) based on the angular spectrum approach is described. The method proposed includes temporal deconvolution of the waveform data to enhance both axial and lateral resolutions, control of the aperture and of the frequency bandwidth to improve signal-to-noise ratio, as well as spatial interpolation of the subsurface images. All the above operations are well adapted to the frequency domain calculations and embedded in the F-SAFT data processing. The aperture control and the spatial interpolation allow also a reduction of sampling requirements to further decrease both inspection and processing times. This method is of particular interest when ultrasound is generated by a laser and detected by either a contact ultrasonic transducer or a laser interferometer.

The invention discloses a GAN based motion blur removing method of an image and a motion blur removing GAN model for the method. The method comprises that the GAN model is designed; the model is trained; and in an application phase, the GAN model comprises a generator and a discriminator, the generator optimizes parameters continuously so that a generated image approaches distribution of a clear image, the discriminator optimizes parameters continuously so that whether the image is from fuzziness removing image distribution or clear image distribution can be discriminated more effectively, thegenerator comprises a down-sampling device and an up-sampling device, the down-sampling device carries out convolution on the image and extract semantic information from the image, and the up-sampling device carries out deconvolution on the image by combining the obtained semantic information with structure information of the image. Thus, motion blur of the image can be removed effectively, and the clear image satisfying perception of humans is obtained.

The invention discloses a pedestrian recognition method based on combination of depth learning and property learning. According to the invention, a convolution neural network containing five implicit strata is constructed. Network training is performed by an anti-convolution method and a concept of property learning is combined. Preferred features obtained from the convolution neural network are input to property classifiers, so that the posterior probability of the property of a sample is obtained. Then by combining with a property class mapping relation, the posterior probability of the class is obtained, so that the class of the sample can be judged. The method is good in detection recognition performance and intrinsic features of an image can be extracted. Besides, since the property has better semantic expression performance than low-stratum features and due to the insensitivity to light and view angles, the algorithm has a good recognition effect.

An image processing device according to an example of the invention comprises a differentiating section which differentiates input image data to generate gradient image data, a gradient domain deconvolution processing section which applies a deconvolution to the gradient image data, the deconvolution performing deblurring corresponding to a prior distribution of an image gradient of an image to generate deconvolved data, and an integrating section which integrates the deconvolved data to generate deblurred image data.

A radioactive emission probe in communication with a position tracking system and the use thereof in a variety of systems and methods of medical imaging and procedures, are provided. Specifically, wide aperture collimation—deconvolution algorithms are provided, for obtaining a high-efficiency, high resolution image of a radioactivity emitting source, by scanning the radioactivity emitting source with a probe of a wide-aperture collimator, and at the same time, monitoring the position of the radioactive emission probe, at very fine time intervals, to obtain the equivalence of fine-aperture collimation. The blurring effect of the wide aperture is then corrected mathematically. Furthermore, an imaging method by depth calculations is provided, based on the attenuation of photons of different energies, which are emitted from the same source, coupled with position monitoring.

The present invention provides a method and apparatus for the detection of an infectious disease or disorder in a fluid, such as a mammalian blood sample, the detection of a specific protein in a urine sample, or the detection of a particle in a plasma. The identification of the particles of interest is enable by taking a transmission spectrum of a test sample in at least a portion of the ultraviolet, visible, near-infrared portion of the spectrum and comparing the spectrum with a standard sample spectrum. From the comparison it is then determined whether the fluid from the test sample contains an particle of interest, and an identity of the particle of interest is determined. Spectroscopic and multiwavelength turbidimetry techniques provide a rapid, inexpensive, and convenient means for diagnosis. The comparison and determination steps may be performed visually or by spectral deconvolution.

A filter assembly adapted to be used with a camera for selectively controlling the light that reaches the camera's aperture. In one embodiment, the filter assembly comprises three filters adapted to be independently moved between a first position wherein they are not in front of the camera's aperture and a second position wherein they are in front of the camera's aperture. The first and second filters are polarizing filters adapted to block portions of visible light. The third filter is an infrared filter adapted to block infrared light. In addition to moving between its first position and its second position, the second filter is also adapted to rotate up to 360 degrees. The image captured by the camera may be improved using a computer implemented image enhancement system that uses one or more of multi-spectral imaging, deconvolution, edge enhancement, and dynamic range translation.

The invention discloses an image classification method based on a confrontation network generated through feature recalibration. The image classification method based on the confrontation network generated through feature recalibration is suitable for the field of machine learning and comprises the steps that to-be-classified image data are input into a confrontation network model for network training; a generator and a discriminator which are constituted by a convolutional network are constructed; random noise is initialized and input into the generator; the random noise is subjected to multilevel deconvolution operation in the generator through the convolutional network, and finally, generated samples are obtained; the generated samples and authentic samples are input into the discriminator; and the input samples are subjected to convolution and pooling operation in the discriminator through the convolutional network, thus a feature graph is obtained, a compressed and activated SENetmodule is imported into an intermediate layer of the convolutional network to calibrate the feature graph, thus the calibrated feature graph is obtained, global average pooling is used, and finally,image data classification is output. The SENet module is imported into the intermediate layer of the discriminator, the importance degree of each feature channel is automatically learned, useful features relevant to a task are extracted, features irrelevant to the task are restrained, and thus semi-supervised learning performance is improved.

The invention discloses an image depth estimation method based on a convolutional neural network. The method comprises the following steps: constructing a convolution-deconvolution pair neural networkmodel, wherein the convolution-deconvolution pair neural network model comprises a plurality of different convolutional layers, a plurality of convolution-deconvolution layer pairs and an activationlayer; selecting a training set, and setting training parameters of the convolution-deconvolution pair neural network model; according to the convolution-deconvolution pair neural network model and the training parameters thereof, training the convolution-deconvolution pair neural network model with minimizing of a loss function being a target to form an image depth estimation neural network model; and inputting an image to be processed to the image depth estimation neural network model and outputting a corresponding depth image. The grayscale value of the depth image obtained through the image depth estimation method based on the convolution-deconvolution pair neural network is more accurate, and the sense of depth of the depth image is higher.

The invention provides a medical image segmentation method of a residual full convolutional neural network based on an attention mechanism. The medical image segmentation method comprises the steps: preprocessing a to-be-segmented medical image; constructing a residual full convolutional neural network based on the attention mechanism, wherein the residual full convolutional neural network comprises a feature map contraction network, an attention network and a feature map expansion network group; inputting the training set data into a residual error type full convolutional neural network for training to obtain a learned convolutional neural network model; and inputting the test set data into the learned convolutional neural network model, and performing image segmentation to obtain segmented images. According to the medical image segmentation method, an attention network is utilized to effectively transmit image features extracted from a feature map contraction network to a feature mapexpansion network; and the problem of lack of image spatial features in an image deconvolution process is solved while the attention network can also inhibit image regions irrelevant to a segmentation target in a low-layer feature image, so that the redundancy of the image is reduced, and meanwhile, the accuracy of image segmentation is also improved.

Object motion during camera exposure often leads to noticeable blurring artifacts. Proper elimination of this blur is challenging because the blur kernel is unknown, varies over the image as a function of object velocity, and destroys high frequencies. In the case of motions along a 1D direction (e.g. horizontal), applicants show that these challenges can be addressed using a camera that moves during the exposure. Through the analysis of motion blur as space-time integration, applicants show that a parabolic integration (corresponding to constant sensor acceleration) leads to motion blur that is not only invariant to object velocity, but preserves image frequency content nearly optimally. That is, static objects are degraded relative to their image from a static camera, but all moving objects within a given range of motions reconstruct well. A single deconvolution kernel can be used to remove blur and create sharp images of scenes with objects moving at different speeds, without requiring any segmentation and without knowledge of the object speeds.