127results about How to "Preserve image details" patented technology

The invention provides a compressed sensing theory-based reconstruction method of a magnetic resonance random sampled K space data image. The reconstruction method applies a contourlet conversion and iterative soft thresholding method to realize reconstruction of a magnetic resonance image. The method comprises the following steps: collecting K space data in a magnetic resonance image scanner according to a preset observation matrix phi to generate a measurement value, and keeping y; acquiring y from a coil of the magnetic resonance image scanner, and transmitting y to a computer; and finallyconstructing a same phi, constructing any orthogonal transformation psi, and recovering from y by adopting a compressed sensing theory-based magnetic resonance random sampled K space data image reconstruction method according to reconstruction. According to the method, scanning time is saved, quick imaging is realized, high-quality reliable image information is provided to medical nuclear magnetic resonance imaging detection, and solid theoretical and practical foundation is established for further development and large-scale popularization and application of the medical imaging detection technology.

The method comprises: reading out a digital image; saving the color and gray values of each pixel into an allocated memory; dividing the inputted image into a light illumination image and a reflection image; respectively making the treatment for both images; after treatment, said two images are combined into one image that is outputted to the output device. By the invention, the digital photos can get a high dynamical range (from the most brightness to the most darkness) according to the partial image information obtained at different light condition.

The invention relates to a motion-estimation-based road video stabilization method, which comprises the following steps of: performing nonlinear smooth denoising on an input image; determining a characteristic searching range according to the requirements of a system on accuracy, and acquiring optimal characteristic information; performing cross grayscale projection on a template and a block to be matched, and finding an optimal matched area according to a difference; constructing a dynamic motion model in a recursion way by adopting a Kalman filter with a uniform motion equation, obtaining motion parameters of a camera for describing inter-frame motions caused by the motions of the camera, and identifying and filtering random jitter in an offset sequence obtained by motion estimation; and filling an area losing an image due to the video jitter, and reconstructing adjacent frames by adopting a hybrid Gaussian model and utilizing time relativity between image sequences.

Disclosed is a blind restoration method for the motion blurred image; the steps are as follows: (1) the image is converted through the cepstrum method to figure out the blur extent and the blur direction of the blurred image; (2) the blur extent and the blur direction of the blurred image figured out in step (1), and the total variation (TV) method are adopted to process the restoration towards the blurred image. The method based on the total variation (TV) includes: the fixed-point iteration used as the outer-loop and the conjugate gradient method used as the inner-loop are adopted for loop iteration to obtain the restored image. The blind restoration method has the advantages of strong self-adaptation, strong anti-noise ability and good robustness; the blind restoration method has validity and practicality according to the processing effects of the simulation picture and the photographed picture.

The invention puts forward a compressed sensing reconstruction method based on image nonlocal similarity. According to the method, image nonlocal similarity, a low-rank matrix and minimum total variation (TV) are combined, and two priors, namely, the local similarity and local smoothness of images, are fully utilized. On one hand, the block effect and the loss of global structure information caused by independent processing of single image blocks in the traditional method are eliminated, and on the other hand, real details of images are retained and false details produced by unreliable information are reduced or removed while noise is suppressed. High-quality compressed sensing reconstruction of images is realized. Compared with a general reconstruction method based on transform domain sparse or TV constraints, the method of the invention is of robustness to noise, better reconstruction quality is achieved, and great improvement is achieved both in visual effect and evaluation index.

The invention discloses a de-noising method of filtering images in size adaptive block matching transform domains. Less false signals are introduced as two-dimension transformation of each image block in the block matching 3D (BM3D) in a basic estimation stage is abandoned by the method; image details can be well preserved as the block number in blocking matching groups of the method is less than the block number in the BM3D method. The image de-noising performance of the method is further improved as the method adaptively selects the block size based on form components during block matching. The current general objective evaluation of image de-nosing includes peak signal noise ratio (PSNR) and mean structural similarity (MSSIM), and according to the method, the de-noising calculation results of a plurality of standard images provided on BM3D networks are higher than the results of the BM3D method on the basis of the two objective evaluations and under all noise intensities.

According to the legend information, in view of the different characteristics of the color, shape, texture and spatial relationship of various elements on a geological map, techniques of color segmentation, boundary tracking, and mode identification are adopted to extract the main information, e.g. stratum, magmatite, construction, or ore occurrence, related to mineral products in the geological map to form a thematic information graph layer. The geological map thematic information extraction method includes a geological map data preprocessing method, a geological map thematic information extraction method, and an information output method. The method aims to solve the difficulty in geological map information extraction, namely, takes a semi-structured standard geological map as a research object and the geological map information extraction as objective, and extracts the main information, e.g. stratum, magmatite, construction, or ore occurrence, related to the mineral products in the geological map through the combination with legend information to construct the thematic information graph layer, thereby forming a set of techniques for converting the semi-structured data (image) into structure data.

In the system, the input module is used for collecting the inputted digital image; the collected digital images is inputted into the image dissection module; the image dissection module dissects image into a light illumination image L and a reflection image R, both are respectively inputted into the light illumination processing module and the reflection processing module; the light illumination processing module makes non-linear correction for the light illumination image L and output an corrected light illumination image L'; the reflection module makes the partial noise reduction process for the reflection image R corresponding to the dark area of the inputted image and outputs a noise-reduced reflection image R'; the image combining and outputting module combines L' and R' and then displays the combined image on the output device.

The invention provides an image noise reduction method and device, electronic equipment and a storage medium. The method comprises the following steps: determining exposure parameters according to a shooting scene, shooting by adopting the exposure parameters to obtain a shot image, determining a matched noise reduction model according to the equipment attribute of imaging equipment for shooting, and inputting the exposure parameters and the shot image into the noise reduction model to perform noise reduction on the shot image, wherein the noise reduction model is learned to obtain a mapping relation between the exposure parameters and the noise characteristics. According to the method, after the matched noise reduction model is determined according to the equipment attribute of the imaging equipment used for shooting, noise reduction processing is carried out on the shot image shot in the current shooting scene, and more image details are reserved while the image purity is ensured. Besides, the shot image is subjected to noise reduction through the noise reduction model matched with the equipment attribute, the noise characteristic can be better identified, the purpose of noise reduction is achieved, the noise reduction effect of the image is improved, and a clearer image can be obtained.

The invention relates to the technical field of brain tumor image processing, and specifically discloses a brain tumor segmentation data enhancement method based on a generative adversarial network. According to the method, an image-to-image conversion framework-Tumor GAN network architecture (comprising a generator, a global discriminator and a local discriminator) is constructed; n2-n virtual image pairs can be synthesized from n real data pairs (a multi-modal brain tumor picture set with segmentation tags), and a data enhancement model is obtained through training, so that the quality and authenticity of an image synthesized by the data enhancement model are ensured; on the basis of an attention region provided by a semantic tag, a region perception loss function and a region loss function are added in a generator, training of an image with authenticity to an image translation model is assisted, image details can be reserved, and the generalization performance of the model is improved; by applying the local discriminator to be used in cooperation with the global discriminator, the discrimination efficiency can be improved, and the model is helped to generate a medical image pairwith more real texture details.

The invention discloses a stripe noise removal method for an infrared imaging system. The method includes the following implementation steps that first, a black body is imaged through the infrared imaging system, and an obtained infrared image is converted into a one-dimensional number sequence; second, FFT is performed on the obtained one-dimensional number sequence; third, a spectral line representing stripe noise in a spectrogram after transformation is calibrated; fourth, the result of the image which is also converted into the one-dimensional number sequence, is about to be denoised and has been subjected to FFT is processed at the position of the calibrated spectral line; fifth, inverse transformation is performed on the processed result so that the one-dimensional number sequence can be obtained, and then the two-dimensional image is formed through splicing. The method overcomes the defects of a traditional infrared image noise removal method, and is more effective and more accurate. Through the method, stripe noise in the infrared image can be effectively removed, loss of original information of the image is avoided, and the high-quality restoration image is obtained.

The invention provides a palm print database search method based on quantum algorithms. Four different quantum algorithms are respectively used in the three most important stages of palm print recognition. The search method comprises the following steps that 1, a palm print image is filtered by a quantum adaptive median filtering algorithm, and a fact can be observed from a filtering effect contrast chart that a better filtering effect can be obtained by the quantum adaptive median filtering algorithm utilized in the invention than a classic algorithm; 2, the characters of the palm print image are extracted by quantum Fourier conversion based on quantum parallelism, all palm print characters can be extracted by only one quantum Fourier conversion operation, and an analysis result shows that the extraction speed of the palm print characters is raised by the algorithm under the condition of well reserving palm print details; and 3, the palm print image is matched by a quantum set operation and a Grover algorithm. In comparison with the classic algorithm which requires N times of operation, the matching algorithm provided by the invention has the advantage that the position of the target palm print image can be found by only 1 / 2 N times of operation, and the probability of finding the position of the target image in the palm print database by the algorithm is close to 1.

The invention provides an image processing method and device based on multiple frames of images. The method comprises the steps that multiple frames of original images are acquired, a neural network model corresponding to each frame of original image is selected according to the exposure degree of each frame of original image for noise reduction processing, so that each frame of original image after noise reduction is obtained, and a target image is synthesized through each frame of original image after noise reduction. Therefore, by performing noise reduction processing on multiple frames oforiginal images, the effectiveness of noise reduction is improved, image details are reserved while image noise is reduced in the synthesized target image, an imaging effect with better definition isobtained, and the imaging effect of the target image is improved.

The invention discloses a time domain high pass non-uniformity correction method based on gray scale correlation, relating to the time domain high pass non-uniformity correction method based on gray scale correlation in the infrared imaging technology field and belonging to the infrared imaging technology field. The time domain high pass non-uniformity correction method based on the gray scale correlation comprises steps of using spatial domain low pass filtering result having edge protection as a correction reference source to perform pre-correction on an inputted image, calculating correction offset value of each frame by combining with time domain high pass filtering, changing the mapping relation between the offset value and the grey scale according to the changing volume of the incident radiation in the same position of each frame, removing the ghost during the correction procedure, and improving the infrared imaging quality. The time domain high pass non-uniformity correction method based on the gray scale correlation can reduce the occurrence probabilities of the ghost and the over-correction of the real-time infrared imaging system non-uniformity correction algorism, improves the infrared imaging quality, reduces the calculation quantity and storage space and facilitates the realization of the hardware.

The invention discloses a magic square coding-based median filter method comprising the following steps of: firstly, reading the coordinates and the gray value of each pixel in an image; secondly, traversing the image and judging whether each pixel is a noise pixel or positioned in a linear detail area or not by using a magic square coding-based method; and thirdly, maintaining the gray value of a non-noise image unchanged and removing the noise of the noise pixel by amending the gray value of the noise pixel. The invention can be used for judging whether the pixel processed as the noise pixel is positioned in the linear detail area or not without generating pixel loss in the linear detail area, thereby ensuring the integrity of image information; the pixels in the linear detail area are judged by using the magic square coding-based method with small calculated quantity; and a proper value is selected to replace the noise pixel according to the difference condition between the gray median value and the mean value of the maximum gray value and the minimum gray value so as to reduce the fuzzy phenomenon caused during the the image noise removing processing.

The invention discloses a CT image denoising method based on self-adaptive median filtering, belongs to the technical field of medical image processing, and is particularly suitable for CT image denoising of new coronal pneumonia. The method comprises the steps of selecting a square filtering window with the size of n * n, comparing the adaptive maximum value and the adaptive minimum value of the gray value in the window with the gray value f (i, j) of the current pixel point, judging whether the current pixel point is a suspected noise point or not according to a first threshold T0, and if so, further accurately judging whether the current pixel point is a noise point or not according to a second threshold T1, if the current pixel point is not the suspected noise point or the noise point, traversing the next pixel point in the window; processing noise points through a center weighted median filtering method; and finally, outputting the CT image after median filtering and denoising. Image details are better protected while image denoising is kept; the problem that deviation exists in information transmission of a traditional entropy weight method is corrected by improving the entropy weight method, the optimal weight of center weighted filtering is determined through contribution values of all evaluation indexes to the denoising effect, and therefore the optimal denoising effect is achieved.

The invention discloses an average filtering method eliminating image impulse noise fast and efficiently. The method comprises the following steps that filtering windows of images are set in the 8-bit grey-scale images, if the grey value of center pixels is 0 or 255, the pixels are suspected to be the noise pixels, and otherwise the pixels are the signal pixels; if the pixels are suspected to be the noise pixels, the signal pixels are extracted from the boundary pixels to form a set, and if the pixels are the signal pixels, the pixels are not processed; if the set formed by the extracted signal pixels is a non-null set, processing is not carried out, and if the set is a null set, two is added to the number of the filtering windows until the new set is a non-null set; the element grey value average of the set is used for replacement, and processing of the above steps is repeated until all suspected noise pixels obtain the filtered images. The method has good abilities to eliminate noise and maintain details, is high in practicality, and achieves adaptive filtering of the images, high-density impulse noise can be eliminated simply and efficiently, the edge and image details are retained, and the efficiency is higher than that of a traditional filtering method.

The invention discloses a salt and pepper noise filtering method based on a similar function, comprising the steps of setting an image filtering window in a 8-bit gray level image, determining noise pixels or suspected noise pixels through central pixel gray level values, and otherwise, judging the pixels to be signal pixels, establishing a marking matrix to mark the noise pixels, eliminating the pixel points in the filtering window, wherein the gray levels of the pixel points are identical to that of the central pixels, constituting a first signal point set by using the signal points which satisfy the marking matrix elements, processing the signals in the first signal point collection to obtain a first output image, setting the marking matrix element corresponding to the position of the output pixel in the output image to finish the filtering processing. The invention has better capability of denoising and maintaining details.

The invention discloses a self-adaptive median filtering denoising method applied to image processing, which comprises the following steps of: firstly preprocessing an image, and then, (1) sorting data in a filtering window by adopting a divide-and-conquer sorting strategy; And (2) adaptively adjusting the size of a filtering window by combining the noise density. The adaptive median filtering denoising method applied to image processing is high in denoising efficiency and easy to implement.

The invention discloses an L1 norm total variation regularization heterogeneity correction method based on anisotropy. The method comprises the following steps that: establishing anisotropy total variation regularization model based on a single image, converting a problem of removing infrared image heterogeneity stripe noise into a minimized total variation problem, then, adopting a separated Bragg iteration method to carry out optimization to obtain an optimal solution, and taking a last iteration result as a corrected infrared image. The method has an innovation point that a traditional total variation model aims at a structural feature improvement equation that the horizontal direction total variation of heterogeneity stripe noise is far greater than vertical direction total variation,a regularized constraint based on the L1 norm can be used to enable the method to be suitable for the heterogeneity correction of the infrared image, the separated Bragg iteration method replaces a steepest gradient descent method to carry out the optimal processing of the equation, processing speed is greatly improved, and the edge information of an object is kept to a maximum degree.

The invention provides an image noise reduction method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a plurality of frames of original images, synthesizing according to the plurality of frames of original images to obtain a high-dynamic-range image, determining a matched noise reduction model according to ambient brightness, and performing noise reduction on the high-dynamic-range image by adopting the noise reduction model to obtain a target noise reduction image. According to the method, the ambient brightness of a plurality of frames of original images is collected, after a matched noise reduction model is determined, noise reduction processing is carried out on a high-dynamic-range image, and more image details are reserved while the image purity is ensured. In addition, noise reduction processing is carried out after multiple frames of original images are synthesized; therefore, both the bright light area and the dark light area in the synthesized high-dynamic-range image can be denoised, the noise reduction effectiveness is improved, image details are reserved while the image noise of the target noise-reduced image obtained through noise reduction is reduced, and an imaging effect with better definition is obtained.

The invention discloses a defective pixel detection and correction method and device, a storage medium and a terminal, and the method comprises the steps: moving a sliding window pixel by pixel in a to-be-detected image, and calculating the detail density corresponding to a central pixel through the pixel value of each pixel in the sliding window; after the sliding window is moved, determining a defective pixel threshold coefficient corresponding to the central pixel at least according to the detail density corresponding to the central pixel; after the sliding window is moved, dividing the sliding window into a plurality of sub-windows, calculating a defective pixel threshold range corresponding to each sub-window according to the pixel value in each sub-window and the defective pixel threshold coefficient corresponding to the central pixel, and comparing whether the pixel value of the central pixel falls into the defective pixel threshold range corresponding to each sub-window or not;and after the sliding window is moved, determining whether the central pixel is a defective pixel or not according to a comparison result of the central pixel and the defective pixel threshold rangecorresponding to each sub-window until the to-be-detected image is traversed. According to the technical scheme, the accuracy of defective pixel detection can be realized.

The invention discloses a removal method of a zipper effect of bayer interpolation. The bayer interpolation is a device of interpolating missing color components of pixel points of a color filter array; and the zipper effect is an image error appearing most easily in the bayer interpolation. According to the characteristics that the zipper effect mutational in a brightness domain and even in a chroma domain, the method comprises the steps of converting RGB (red, green and blue) image data after the bayer interpolation to a brightness-chroma color space (such as YUV and Lab) via color space transformation, determining texture types of regions where pixel points of an image brightness component are located one by one, performing low-pass filtration in case of an even region, performing horizontal and vertical filtration in case of a horizontal or vertical marginal region, constituting a new image with the filtered image brightness component and the original chroma component, and performing color space inverse transformation to form a final RGB image. The method can effectively eliminate the zipper effect generated by the bayer interpolation and significantly improves image quality.

The invention discloses a sharpness processing method for foggy-day images. The sharpness processing method for foggy-day images comprises utilizing different scales of Gaussian kernel function Fi(x,y) to carry out convolution for an original foggy-day image to obtain a brightness image Li(x,y) which can reflect the ambient brightness; summing the logarithmic difference of the original image V(x,y) and the brightness images Li(x,y) according to the weight so as to obtain a reflection image R(x,y); performing contrast stretching for the obtained reflection image R(x,y) to obtain a stretched reflection image R'(x,y); using improved gamma correction to correct the R'(x,y), wherein the correction function is Ia(a+I); and at last selecting two platform thresholds T1 and T2 to perform double-platform histogram processing for the image after gamma correction to obtain an image after the final processing. The algorithm consummates the deficiency of a Retinex algorithm, and gets good balance between the image dynamic range and the detail fidelity, and not only improves the contrast but also preferably keeps the details of the image.

The embodiments provide an image processing method and apparatus, which combine a wavelet transform method with a non-local means method, comprising: performing N-level decomposition of an original gray image by using a wavelet transform method; and performing non-local means processing and wavelet reconstruction of each level of components one by one starting from an N-th level, so as to obtain a processed gray image. The image processing method and apparatus can effectively remove various image noises, and well reserve the image details.

The invention discloses an SAR (synthetic aperture radar) despeckle method based on target extraction and three-dimensional block matching denoising, and mainly solves the problem that despeckle effect and detail keeping cannot be considered at one time in the prior art. The implementation steps of the method include: firstly, extracting and removing a target image of an input SAR image; secondly, calculating an iterative initial value, and setting iterative times and iterative step size; thirdly, performing logarithmic transformation to an image after the target image is removed, denoising by the three-dimensional matching denoising algorithm with different standard deviations of noise, and selecting the best noise deviation by evaluating the edge preserving capacity and equivalent number of looks of the denoised image; fourthly, performing three-dimensional matching denoising to a logarithmic image by the best standard deviation of the image; and fifthly, performing exponent transformation to the denoised image, and adding the image with the target image to acquire a final despeckled image. The method has the advantage that image details are saved well, is effective in despeckle, and can be used for SAR image target identification and target detection.

The invention belongs to the technical field of image processing, and particularly relates to a low-dose CT image denoising recursion algorithm based on a residual encoding and decoding network. According to the specific technical scheme, the invention discloses a shallow residual error coding and decoding recursive network for low-dose CT image denoising. The recursive shallow residual encoding and decoding network reduces the complexity of the network by reducing the number of layers and the number of convolution kernels in the residual encoding and decoding network. The network performanceis improved by using a recursion process. According to the algorithm, end-to-end mapping is trained through a network to obtain a high-quality image. At each recursion, original low-dose CT images arecascaded to the next input. The problem of distortion of the image after multiple recursions can be effectively avoided, image features can be better extracted, detail information of the image can bereserved, the complexity of the network can be reduced, the network performance can be improved, image details of the denoised image can be well reserved, and the image structure is clearer.

The invention provides an edge detection method of a Canny operator based on high and low thresholds, which mainly comprises the following steps of: 1) carrying out smooth filtering processing on a source image, and removing noise of the image by using switch median filtering instead of Gaussian filtering; (2) calculating the gradient magnitude and direction of the image after smoothing filtering and denoising in the step (1) by adopting a sobel operator; 3) performing non-maximum suppression on the gradient magnitude obtained in the step 2) to obtain an edge image with a single-pixel width; 4) adopting a k-means clustering algorithm to obtain clustering centers of high and low gradient values; 5) obtaining an otsu threshold value of the gradient through an otsu algorithm, and taking a high threshold value and a low threshold value among the high clustering center, the otsu threshold value and the low clustering center; 6) processing the edge image of the single pixel width obtained in the step 3 by using a high threshold value and a low threshold value, and obtaining a binary edge; and 7) removing the interference edge of the binarized edge by adopting an area morphological opening operation, and obtaining a final edge image. The Canny algorithm provided by the invention has the advantages of high positioning precision, strong adaptability, good interference point removal effect and the like.

An image processing device for reducing image noise and the method thereof are provided. The method for reducing image noise includes the following steps. A filtering operation is performed on a plurality of pixel values of a first image according to a high-pass (HP) filtering mask, a first default filtering mask and a second default filtering mask to obtain a high-pass filtering image, a first low-pass (LP) image and a second LP image, respectively. A difference between the first and the second LP images is calculated to obtain a LP difference image. A pixel weighting calculation is performed according to the HP image and the LP difference image to obtain a pixel weighting mask corresponding to the first image. The first image is calculated according to the pixel weighting mask and one of the first and the second LP images to obtain a noise reducing image.

The invention discloses a pyramid-based image filtering method and device, and a storage medium. The method comprises the steps: constructing an image pyramid of a to-be-processed image; sequentiallytaking down-sampled images from the image pyramid according to the sequence of the resolutions to serve as current layer images; filtering the current layer image to obtain an intermediate filtered image; performing up-sampling on the intermediate filtered image to obtain an up-sampled image; fusing the upsampling image with the next layer of image of the current layer of image to obtain a fused image; taking the fused image as a current layer image, and judging whether the current layer is the last layer or not; if the current layer is not the last layer, returning to the step of filtering the image of the current layer to obtain an intermediate filtered image; and if the current layer is the last layer, obtaining an intermediate processing image, fusing the intermediate processing imagewith the to-be-processed image, filtering the fused image to obtain a filtered image, and outputting the filtered image. In this way, errors can be suppressed, and the filtering effect is improved.