397 results about "Image recovery" patented technology

The invention provides a camera module. The camera module comprises a first lens array, a second lens array, a liquid crystal shutter arranged between the first lens array and the second lens array, a controller, an image sensor and an image processor, wherein the first lens array comprises a plurality of first lens; the second lens array comprises a plurality of second lens which are respectively in optical alignment with the plurality of first lens one by one; the liquid crystal shutter is provided with a plurality of light-through holes; the plurality of transmission holes are respectively in optical alignment with the plurality of the first lens one by one; the controller is used for controlling every light-through hole of the liquid crystal shutter to open once in a shooting period; the image sensor is used for obtaining a sub-image corresponding to each light-through hole when each light-through hole in the shooting period is opened; and the image processor is used for synthesizing each sub-image corresponding to each light-through hole to an image by using a super-resolution image recovery technology.

A method and apparatus to recover a set of data from a full backup and an incremental or differential backup are described. This includes a source directory that was modified between the full backup and the incremental or differential backup. The full backup is restored, including the source directory, and a new directory is created to replace the restored directory when the incremental or differential backup is applied to the restored full backup. Content for an entry from the modified source directory in the incremental or differential backup is created in the new directory if the corresponding content is present in the incremental or differential backup. If the content for the entry is not present in the incremental or differential backup, then the entry in the new directory is linked to corresponding content in the restored directory from the full backup.

An embodiment of the present invention is a system and method relating to automatic firmware image recovery. When a server equipped with a baseboard management controller (BMC) and operational code detects that its operational code image is corrupted or out of date, it broadcasts a request for an image update over an out-of-band network. One or more donor systems on the network may respond to the request and send the requestor a new image. The recipient system use management policies to determine from which donor system to accept an update.

The invention discloses a method for hiding high-capacity compression-resisting image information. According to the method, firstly, a carrier image is sampled and is decomposed to acquire sub-images which are same in size and do not mutually overlap; secondly, several sub-images are preferentially selected from the sub-images and serve as carrier sub-images, and the rest of the sub-images serve as source images used for carrier recovery. Firstly, encoding and scrambling are conducted on embedded information; secondly, one or more digits among four high digits in the carrier sub-images are hidden with a foundational hiding method, and sub-images which contain secret information and sub-images which do not contain secrete information are reconstructed to form an image which contains secret information to be transmitted; received images are decomposed by a receiving end, the secret information is extracted from the sub-images which contain the secret information, and the sub-images which do not contain the secret information are predicted in an interpolation mode to recover carrier images. The method has the advantages of being large in hiding capacity, strong in compression-resisting capability and low in complexity, can recover the secret information and the carrier images in a high-quality mode under the condition of achieving two-time compression to eight-time compression of a JPEG2000 algorithm, wherein the maximum relative hiding capacity can reach one eighth and the typical value of the PSNR of carrier image recovery is 40dB.

The invention provides a deep learning neural network-based microscopic image three-dimensional reconstruction method and system. The method comprises the following steps of constructing a neural network; obtaining a training set of the neural network; training the neural network according to the training set, thereby obtaining network parameters; and according to the network parameters, performing three-dimensional reconstruction on a to-be-reconstructed object to obtain a reconstructed image. According to the method and the system, an image recovery reconstruction network is obtained through learning of an optical field image and different layers of focus images, so that the three-dimensional reconstruction speed and resolution are increased, and the longitudinal resolution is greatly increased.

The invention discloses a depth image recovery method based on improved bilateral filters. The method includes the steps that firstly, the local self-adaption bilateral filter is used for carrying out smooth denoising on an original depth image; then error depth pixels in the depth image are corrected; finally, an RGB image corresponding to the depth image is combined, holes of the depth image are filled through the selective combined bilateral filter, the noise of the finally recovered depth image is effectively restrained, the error depth image pixels are effectively corrected, and a clear depth boundary is provided. The local nonuniform noise in the depth image is effectively restrained, the holes in the depth image are filled, the edges of the recovered depth image are clear and orderly, and later three-dimensional scene recognition, reconstruction and the like are facilitated.

The invention relates to a binocular speckle structured light three-dimensional reconstruction method and system. The method comprises the steps that a speckle image is acquired; building a binocularstructured light vision system, and calibrating to obtain internal parameters and external parameters of the left and right cameras; projecting the speckle image to the surface of the measured objectthrough projection equipment, and shooting through a left camera and a right camera to obtain a left image and a right image; epipolar correction is performed on the left and right images; performingimage matching on the left image and the right image by adopting an improved SGM algorithm based on the speckle image to obtain a parallax image; and based on the internal parameters and the externalparameters of the left and right cameras and the parallax image recovery depth information, obtaining three-dimensional data of the surface of the measured object, and obtaining a final image. According to the method, the speckle distribution constraint window is preset to control the density degree of pseudo-random speckle distribution, the speckle pattern with high identification degree and richtextures is generated, and the new parallax calculation cost function is established, so that the image matching precision is effectively improved, and the accurate three-dimensional dynamic reconstruction method is realized.

A face portion of an input image, an example of a predetermined structure, is applied to a mathematical model by the image recovery/addition section to recover a missing element of the face portion in the input image. The mathematical model is generated by a predetermined statistical method, such as the AAM scheme or the like, based on a plurality of sample images representing the face portion including the recovery target element. Thereafter, the face portion is reconstructed to include the missing element based on the parameter corresponding to the face portion obtained by applying the face portion to the model, and the face portion of the input image is replaced by the reconstructed face portion to produce a restored image by the image reconstruction section.

The invention relates to a green component and color difference space-based Bayer format color interpolation method. Two main errors which are jaggies and pseudo colors respectively exist in the conventional color interpolation methods. In the method, in a full-color image recovery process, color recovery is realized by adopting a combined way of interpolating green components, red components and blue components in steps; and the method comprises the reconstruction of the full-resolution green components G, the calculation of pixel point color differences R-G or B-G comprising the red or blue components, the reconstruction of full-resolution color difference images R-G and B-G, and the recovery of the red components R and the blue components B by the addition of the reconstructed full-resolution green components and the reconstructed color difference images. The method of the invention has the advantages of effectively solving the problem that the images have blurry edges and unsharp detail textures, reducing the color distortion of the images, preventing color mutation and improving color smoothness.

An image recovery algorithm that recovers completely lost blocks in an image/video frame using spatial information surrounding these blocks. One application focuses on lost regions of pixels containing textures, edges and other image features that pose problems for other recovery and error concealment algorithms. The algorithm of this invention is based on the iterative application of a generic de-noising algorithm and does not require any complex preconditioning, segmentation, or edge detection steps. Utilizing locally sparse linear transforms and overcomplete de-noising, good PSNR performance is obtained in the recovery of such regions.

An image processing apparatus which enables to suppress coloring in an image after image recovery processing, and a control method for the image processing apparatus are disclosed. After the chromatic difference of magnification of an optical imaging system is corrected, a first value relating to coloring is obtained before application of an image recovery filter. Then, an image recovery filter which, preferably, does not correct a chromatic difference of magnification and is based on a function indicating the aberrations of the optical imaging system is applied. A second value associated with coloring is obtained from the image after application of the image recovery filter. According to the difference between the first and second values obtained before and after application of the image recovery filter respectively, pixel value correction for correcting coloring is performed.

The invention discloses an improved image enhancement method based on block matching and recovery and combined trilateral steering filtering. With the method, removing of lots of holes and severe noises in a collected depth image is realized. The method comprises: pretreatment of an original image is carried out; a to-be-recovered unknown region in the image after pretreatment is marked; a pixel point priority level is calculated; a to-be-recovered block is selected; an optimal matching block is searched; recovering is carried out; determination is carried out; the image is processed by using combined trilateral steering filtering; and image enhancement processing is ended and a result is outputted. According to the method provided by the invention, with introduction of a horizontal set distance factor and a diffusion time function, a block-matching-based image recovery sequence is improved and a structural hole is filled; and on the basis of the combined trilateral steering filtering, a pseudo texture generated by recovering and lots of original noises are removed. In terms of a visual effect and a quantitative analysis, the texture structure of the original image can be kept well for the image after enhancement processing and noise information in the image is removed, so that the image distortion degree is low.

Provided is an image quick defogging optimized method based on a black channel. Edges of an image and non-edge areas of the image are processed through different templates to obtain a transmission picture, and a precise atmosphere light value is obtained by the fact that a sky area or an area with thickest fog is divided. The method greatly reduces consumption, improves the image recovery speed, and improves image recovery effects.

The invention relates to a RGB-D camera depth image recovery method combining color images, and belongs to the field of depth image recovery. The method comprises that a Zhang's calibration method isadopted to calibrate color and depth cameras, and internal and external parameters of the cameras are acquired; according to the principle of pin-hole imaging and coordinate transformation, coordinatealignment of the color and depth cameras is achieved; color and depth images of a scene are respectively acquired, and a depth threshold method is adopted to carry out binaryzation processing on theimages; a size of a cavity connected domain is judged to determine existence of cavities; an expansion operation is carried out to obtain a cavity neighborhood; depth value variances of pixel points of a cavity domain are calculated, according to the depth value variances, the cavities are divided into shielding cavities and in-plane cavities; color consistency and domain similarity are respectively adopted to recover the two types of cavities; a partial filter method is adopted to filter the recovered images, and noises are removed. The method ensures the depth image recovery, keeps the original information of the depth images as much as possible, and improves the recovery efficiency of the depth images.

An image processing apparatus and image processing method, which enable suppression of coloring that occurs in image recovery processing for a RAW image are disclosed. Pixel interpolation processing is executed for a predetermined reference color component of the RAW image. For each color component other than the reference color component, a first color difference is obtained using a pixel value of the color component and a corresponding pixel value of the reference color component after the pixel interpolation processing. After applying the image recovery processing to each color component, a second color difference is obtained in a similar manner to the first color difference, and a pixel value is corrected according to the difference between the first and second color differences determined before and after the image recovery processing.

Structures and methods thereof for a scanner with two CCD arrays, which respectively capture two independent object images in one scan action, are provided. Three functions: a. image recovery, b. double resolution, and c. high dynamic range can be achieved for the transparent media and the reflective media. The image defects caused by the dust, pollutants, or scratches can be recovered by comparing the different shadows formed by two light-sources in two object images for the reflective media or comparing two object images formed respectively by a visible light-source and a infrared light-source for the transparent and reflective media. The function of double resolution can be achieved by making a relative displacement of half pixel-distance between two CCD arrays in their longitudinal direction. The function of high dynamic range can be achieved by setting two different exposure time for two CCD arrays.

The invention discloses an adaptive difference expansion-based reversible image watermarking method, which comprises a watermark embedding process, a watermark extraction process and an image recovery process; before watermarks are embedded, the original image needs to be pretreated, and the pretreatment comprises alternated division, difference calculation, image complexity calculation and parameter setting; in the embedding process, the corresponding watermarks are embedded by using adaptive expansion; additional information is embedded, and the watermark data and the additional information are mixed and embedded under a specific condition; in the extraction process, the additional information is first extracted, and the difference is calculated; the inverse operation of the adaptive difference expansion is performed, the watermark data are extracted, and the original difference is reduced; and finally, the image is reduced through the reduced difference. The method has reversibility, and can provide higher embedding capacity and better image quality, particularly under the condition of many embedded watermark data.

The invention discloses a perfect information non-local constraint total variation method for image recovery. By the method, the problems that edges cannot be sharpened and high-frequency details cannot be recovered during the image recovery in the prior art are solved. The technical scheme is as follows: (1) suppressing noise of an indefinite image by using a non-local mean value filter method; (2) initializing a recovered result by using a wiener filter method; (3) calculating a perfect information non-local weight coefficient matrix; (4) updating the recovered result by using a threshold value iterative formula; (5) suppressing noise of the recovered result by using a total variation denoising method; (6) judging whether to update the perfect information non-local weight coefficient matrix, if so, returning to the step (3), otherwise, executing step (7); and (7) judging whether a stop condition is met, if so, obtaining a final result, otherwise, returning to the step (4) until the stop condition is met. During recovery, the edge of an image can be sharpened and the high-frequency details of the image can be recovered; and the method can be used for recovering the indefinite image in a known indefinite type.

The invention relates to a processing method for correcting the deviation of an image, comprises four steps of mathematical modeling, correction of distortional function parameters, calculation of reverse mapping coordinates and image recovery; the processing method further comprises the following steps of: according to the optical path aberration characteristic of the shooting lens, obtaining a distortional curve based on image coordinates through conversion; correcting the parameters through fitting method; calculating the mapping relation between practical imaging coordinates and ideal imaging coordinates; converting the coordinate system of the mapping relation into an image coordinate system; searching for the reverse coordinate points for the distortional image; and recovering and outputting a non-distortional image according to the image interpolation. In the invention, the complex aberration characteristic is analyzed and converted into the nonlinear curve-fitting question, and the image is zoomed while the deviation of the image is corrected; therefore, the video deviation in real time for images with different sizes is realized, and the errors of coordinate origin generated by the different manufacturing processes of different shooting modules can be corrected without needing to re-calculate the correction parameters; and the processing method is suitable for correcting images or videos with high resolution ratio, and the speed and precision of deviation correction are completely improved.

This invention discloses an image resuming method including: obtaining position registration of updated data of a first storage device registering the updated states of data at different positions, comparing the position registration of the second device with that of the first to get different data positions of them then copying data corresponding to the positions to the first device. This invention also provides a storage device, which mainly compares the position registration for updating data recorded by a registration unit with that of another storage device obtained by the obtaining unit to get the positions itself and another storage device and copies the data corresponding to these positions onto another storage device. This invention also provides a network system.

The invention discloses an image deblurring method based on a multi-task CNN. The image deblurring method comprises the following steps: (1) acquiring a total training set, a test set and preprocessing; (2) performing multi-scale zooming and parameter setting on the image; and (3) deblurring based on a multi-task convolutional neural network, wherein the multi-task convolutional neural network comprises three scales, and each scale is divided into an image deblurring module, an image detail recovery module and a feature fusion module; the image deblurring module comprises an image encoding block E and an image decoding block D, and the image encoding block E extracts image features and encodes the image features, and then a deblurred image is obtained through the image decoding block D; the image detail recovery module only comprises a convolution feature extraction layer with a residual structure, and the size of a network feature map is kept the same as the input size, and only blurred image high-order features are extracted and fused, and high-frequency information is provided for the final image recovery process; and the feature fusion module comprises an image merging module and a convolution layer.

The invention discloses a cGAN algorithm-based super-resolution image recovery technology, and relates to super-resolution recovery for deep convolutional antagonistic neural network images. The existing method needs to carry out a lot of training, the network layers are relatively shallow and the feature weight values included in models obtained through the training are not complete enough, so that the final recovery effect is not ideal. Greyscale maps are used during the training, so that the super-resolution recovery effect for three-channel colored images is not good. Aiming at the defects in the prior art, the invention provides a cGAN algorithm-based super-resolution image recovery method which is capable of training the colored images to achieve better recovery effect. The key points of the method comprises that (1) the training time is short, (2) three-channel colored maps can be directly trained, (3) much image preprocessing is not required, and (4) the models obtained through training can be used once for all.

The invention relates to an underwater image restoration method based on a cyclic generative adversarial network, and the method builds a cyclic generative adversarial network (CycleGAN-VGG) based onperception loss, the network is an end-to-end network architecture, and a restored image is directly obtained through an inputted distorted underwater image. According to the invention, the cyclic consistency loss is reduced; vGG perception loss is added to enhance the structure of the CycleGAN, the overall structure of the input image is reserved by optimizing the cyclic consistency loss, and thedifference of the image feature space is compared by perception loss, so that the network can better recover the detail information of the underwater image, perform underwater image recovery and increase the definition of the image. According to the method, the framework of the Wasserstein GAN is adopted, so that the robustness of the method is improved. Meanwhile, in the training and testing stages, the method does not need paired distortion and real image samples, and does not need any underwater image imaging model parameters.

The invention discloses a face image recovery method and device based on a cyclic neural network, and the method comprises the steps: obtaining a low-resolution image, carrying out the pedestrian facedetection of the low-resolution image, and obtaining a to-be-restored face image; inputting the to-be-restored face image into a pre-built and trained neural network model, and extracting a feature map of the to-be-restored face image according to a preset pixel size; carrying out the nonlinear mapping of the extracted feature map, and obtaining an optimized feature map; carrying out the high-dimension reconstruction of the to-be-restored face image according to the optimized feature map, and outputting a corresponding high-resolution face image. According to the invention, the bidirectionalcyclic neural network technology and the generative adversarial network technology are combined for constructing and training a neural network model with the optimal recovery performance, so the method can achieve the effective recovery of a high-resolution face image from the low-resolution face image.

The invention discloses a low-illumination color image enhancement method based on Retinex and a convolutional neural network. The low-illumination color image enhancement method comprises the steps:inputting a low-illumination color image into a decomposition network, and outputting a three-channel reflection image and a single-channel illumination image; inputting the reflection image and the illumination image into a reflection image recovery network, and carrying out denoising and color recovery processing to obtain a recovered reflection image; inputting the illumination image and the illumination adjustment parameters into an illumination image adjustment network, and outputting the adjusted illumination image; and finally, carrying out dot multiplication operation on the recoveredreflection image and the adjusted illumination image to obtain an enhanced image. Based on the Retinex theory, the low-illumination color image enhancement method utilizes the convolutional neural network to realize enhancement of the low-illumination image, constructs a loss function to perform constraint optimization on parameters of the convolutional neural network, so as to achieve the expected effects of low-illumination image brightness, contrast enhancement and image impression enhancement, remove the influence of noise and color distortion to a great extent, and allow a user to autonomously adjust and enhance the brightness.

The invention provides a multi-temporal data based remote sensing image weighted regression recovery method. If effective data of a remote sensing image fail to be obtained due to data loss, cloud pollution and other causes, the effective data can be recovered through supplementary information contained in data obtained at another time. Firstly, geometric registration is performed on the damaged image and an auxiliary image used for providing reference information; and then pixels similar to a pixel in each area to be recovered are searched in the auxiliary image, and a recovered pixel is obtained after weighted regression calculation of each point to be recovered is performed according to the similar pixel changing relationship between the image to be recovered and the auxiliary image. The corresponding relationship of image similar points in different temporal images is fully used for the multi-temporal data based remote sensing image weighted regression recovery method, weighted calculations of the pixel space and spectrum differences are performed, and the recovered result is enabled to be more close to real data. The calculation efficiency is relatively high and easy to achieve, the practicability is high, and the multi-temporal data based remote sensing image weighted regression recovery method can be used for the business-oriented operation of remote-sensing image recovery.