112 results about "Laplacian pyramid" patented technology

A computer implemented method for fusing images taken by a plurality of cameras is disclosed, comprising the steps of: receiving a plurality of images of the same scene taken by the plurality of cameras; generating Laplacian pyramid images for each source image of the plurality of images; applying contrast normalization to the Laplacian pyramids images; performing pixel-level fusion on the Laplacian pyramid images based on a local salience measure that reduces aliasing artifacts to produce one salience-selected Laplacian pyramid image for each pyramid level; and combining the salience-selected Laplacian pyramid images into a fused image. Applying contrast normalization further comprises, for each Laplacian image at a given level: obtaining an energy image from the Laplacian image; determining a gain factor that is based on at least the energy image and a target contrast; and multiplying the Laplacian image by a gain factor to produce a normalized Laplacian image.

A multimedia compression system for generating frame rate scaleable data in the case of universally scaleable data. Universally scaleable data is scaleable across all of the relevant characteristics of the data (e.g., frame rate, resolution, and quality for video). The scaleable data generated by the compression system includes multiple additive layers for each characteristic across which the data is scaleable. For video, the frame rate layers are additive temporal layers, the resolution layers are additive base and enhancement layers, and the quality layers are additive index planes of embedded codes. Various techniques can be used for generating these layers (e.g., Laplacian pyramid decomposition or wavelet decomposition for generating the resolution layers; tree structured vector quantization or tree structured scalar quantization for generating the quality layers). The system further provides for embedded inter-frame compression in the context of frame rate scalability, and non-redundant layered multicast network delivery of the scaleable data.

The invention discloses an image registration and multi-resolution fusion-based panoramic image splicing method. An adaptive threshold of sequence images is subjected to binary processing and image contour information is extracted by utilizing a pixel neighborhood relationship; a similar region of an image contour structure is quickly estimated through a similarity measurement method for Hu momentinvariant features; and SIFT feature point extraction is performed in a similar region part of two original images. Moreover, feature point initial matching is performed by using a bidirectional K-nearest classification algorithm; mismatched points are screened by an RANSAC algorithm; and an affine transformation matrix among the images is estimated. In addition, in an image fusion aspect, brightness equalization and local color mapping processing is adopted. Then, through an improved Laplacian pyramid multi-resolution fusion algorithm, spliced images are fused to realize seamless splicing. According to the method, the image splicing time is shortened, the image splicing efficiency is improved, the similar region can be quickly located, the mismatch in splicing is effectively improved, and the image registration precision is improved.

The invention discloses a double Laplacian pyramid and convolutional neural network-based image amplification method. The method comprises the steps of obtaining a low-resolution image, and performingbicubic amplification processing to obtain a blurred high-resolution image with a corresponding amplification factor; performing feature extraction processing on the blurred high-resolution image toobtain image features of different scales from large to small; performing feature extraction processing on a low-resolution image to obtain image features of different scales from small to large in layers; and fusing the obtained image features of the different scales from large to small and image features of the different scales from small to large in the layers by reconstructing a super-resolution network to obtain a high-resolution image. By implementing the method, the high-quality high-resolution image can be reconstructed more quickly; the image quality is improved by reconstructing thelow-resolution image; and therefore, the method can be better applied to more scenes.

The present invention is to provide an image processing apparatus using the difference among scaled images as a layered image and a method thereof, which utilize the Gaussian and Laplacian pyramid theory to convert an original image into a plurality of scaled images of different scales, and the difference among scaled images of two adjacent different scales as a layered image of the corresponding layer, so that the edge and line characteristics of a scene of the original image for each layered image can be displayed in different levels sequentially from a clear level to a vague level, and provide a layered image display interface and an image characteristic editing interface for users to examine each layered image through the layered image display interface and edit or perform special effect to each layered image, so as to simulate different visual effects based on different vision models.

The invention discloses an X-ray image enhancement method based on an image pyramid. The method mainly comprises the steps of image segmentation, Gaussian pyramid and Laplacian pyramid construction, image detail enhancement, image reconstruction fusion, background restoration and the like. The method has the advantages that on the premise that the background of the X-ray image is not changed, thecontrast ratio of a black area with small image gray value difference can be remarkably improved, and the outline of an object in the image can be clearly seen.

The invention relates to an image enhancement method, a terminal device and a storage medium. The method comprises: starting from incident component estimation, performing convolution operation on animproved bilateral filtering function and an input image to estimate an incident light component, so that edge signals of the image are reserved, and a filtering effect is achieved; performing downsampling by using a Gaussian-Laplace pyramid to obtain images of different scales of the filtered image, performing image enhancement operation on the images of different scales respectively, and finallyperforming reconstruction through the Gaussian-Laplace pyramid to obtain a final enhanced image. According to the invention, the image contrast can be effectively enhanced, and the color of the processed image conforms to the human eye observation effect.

The invention provides a tropical cyclone intensity objective determination method based on a satellite cloud chart and an RVM. The method is used for constructing a tropical cyclone (TC) intensity objective determination model based on the satellite cloud chart and the RVM (relevance vector machine). The method mainly comprises the following two aspects: 1) carrying out fusion on infrared and water vapor channel cloud charts by utilizing a Laplacian pyramid algorithm, constructing a deviation angle-gradient co-occurrence matrix with the TC center as a reference point, constructing characteristic factors closely related to TC intensity by utilizing a plurality of statistical parameters in the co-occurrence matrix and information of TC kernel scale and center latitude and the like, and establishing the TC intensity objective determination model by utilizing the RVM; and 2) based on the fused satellite cloud chart, and with each point as the reference point in sequence, constructing a deviation angle-gradient co-occurrence matrix and calculating a minimum value, a median value and a mean value of a co-occurrence matrix statistical parameter array. The method constructs the characteristic factors closely related to TC intensity by utilizing the plurality of statistical parameters of the co-occurrence matrix parameter array and information of TC kernel scale and center latitude and the like, and establishes the TC intensity objective determination model by utilizing the RVM.

An embodiment of the invention discloses an image processing method, intelligent equipment and a computer readable storage medium. The image processing method comprises the steps of: acquiring an original image to be subjected to moire processing; and running an image processing model to carry out moire removal processing on the original image to obtain a target image, wherein the image processingmodel is a network model which is obtained by training in advance according to a moire training data set and can remove moire included in the image. The image processing model comprises a multi-bandmodule, wherein the multi-band module is used for processing the original image to obtain N layers of Laplace pyramids about the original image, the multi-band module obtains a first processing resultfeature map based on feature maps corresponding to N spatial frequency bands of N layers of Laplacian pyramids, the target image is obtained according to the first processing result feature map, andN is a positive integer greater than or equal to 2. By adopting the image processing method, the intelligent equipment and the computer readable storage medium, a better moire removing effect can be achieved.

The invention discloses an image super-resolution reconstruction method based on a multi-scale pyramid network. The image super-resolution reconstruction method comprises the following steps of S1, performing shallow feature extraction on an input image; S2, performing feature fusion and feature enhancement on the shallow features through K multi-scale residual modules to obtain richer deep features; S3, performing up-sampling on the deep features by using transposed convolution; S4, reconstructing the image by using residual learning; and S5, taking the reconstructed image as the output of the current pyramid network and the input of the next layer of pyramid network, and continuing to adopt the steps S1-S4 for training, so as to obtain an image with higher resolution. According to the method, a multi-scale residual module is adopted to fuse features to obtain richer features; meanwhile, a Laplacian pyramid network is adopted to gradually upsample and reconstruct a high-resolution image; and by means of the method, images with richer details and higher quality can be reconstructed.

The invention discloses an underwater image splicing method and device, and belongs to the technical field of image processing. The method comprises the following steps: acquiring at least two underwater images to be spliced; sequentially performing red channel attenuation compensation processing and white balance processing on each underwater image to obtain a preprocessed underwater image corresponding to each underwater image; performing image enhancement processing on each preprocessed underwater image based on the multilayer image pyramid to obtain an enhanced underwater image corresponding to each preprocessed underwater image; performing image registration processing on the at least two enhanced underwater images based on grid optimization to obtain registered underwater images; performing image fusion processing on the registered underwater images based on a Laplace pyramid multi-scale image fusion algorithm to obtain a fused underwater image; and determining that the fused underwater image is a spliced underwater image, and outputting the spliced underwater image. The invention realizes underwater image splicing method.

The invention discloses a fire-fighting unmanned aerial vehicle image fusion method based on visible light and infrared thermal imaging. The method comprises the steps: using a visible light camera and an infrared thermal imaging camera of a fire-fighting unmanned aerial vehicle to capture a visible light image and an infrared thermal image respectively; converting the two pictures into an image with the size of 1280 * 720 through a size function in the OpenCV; extracting ROI vertex coordinates in the infrared thermogram and displaying the ROI vertex coordinates in the visible light image by using a red rectangular frame; fusing an image in the edge obtained in the infrared thermogram and an image of a coordinate corresponding to the visible light by using a Laplace pyramid algorithm to obtain fire point information; and after the fire point is detected, calculating the three-dimensional position of the target according to the visual information, and accurately estimating the three-dimensional position information of the fire point under the coordinate system of the fire-fighting unmanned aerial vehicle by calculating the relative height and yaw angle between the unmanned aerial vehicle and the target. The method has the advantages of being simple in structure, convenient to operate and easy to implement.

The invention discloses a vehicle type automatic recognition method and system. The method comprises: obtaining a vehicle roof characteristic acquired by a laser radar; obtaining a first image acquired by a first camera; obtaining a second image acquired by a second camera; determining a transformation matrix between the first image and the second image; splicing the first image and the second image according to the transformation matrix; performing image fusion on the spliced images by employing an image fusion algorithm based on a Laplacian pyramid; performing smooth processing on the fusion image; extracting a vehicle chassis characteristic according to the smoothed image; extracting a vehicle wheel shaft characteristic according to the smoothed image; and determining the vehicle type according to the vehicle chassis characteristic, the vehicle wheel shaft characteristic and the vehicle roof characteristic. According to the vehicle type automatic recognition method and system, the requirement of the installation precision is low, the calculation amount is small, and the accuracy of vehicle type recognition is high.

The invention relates to a remote sensing image space-time fusion method and system. The method comprises the following steps: resampling a high-resolution image in a known period to twice the spatial resolution of a low-resolution image in the known period, and constructing a Laplacian pyramid network comprising two layers; based on a Laplacian pyramid network, determining a sudden change prediction result and a regression coefficient; applying the regression coefficient to the high-resolution image in the known period to obtain a preliminary time change prediction result; correcting the preliminary time change prediction result by using similar pixels in a neighborhood, and determining a gradual change prediction result; performing weighted average processing on the sudden change prediction result and the gradual change prediction result by using a linear weighting method based on a sliding window, and determining a final time-space fusion result, wherein the final space-time fusion result is a high-resolution image in the prediction period.

The invention discloses a self-adaptive exposure adjustment multi-scale entropy fusion underwater image enhancement method, which comprises the steps of applying generalized bounded logarithm additionor subtraction operation to an original underwater image to derive derivative images with different exposures; calculating multi-scale Gaussian pyramid images with different exposure quantities and multi-scale Laplace pyramid images with different exposure quantities; calculating multi-scale Gaussian pyramid weight images with different exposures; obtaining a normalized multi-scale Gaussian pyramid weight iteration image with different exposure quantities; obtaining a multi-scale pyramid fusion image; obtaining a pyramid fusion image; and obtaining an enhanced image based on rich gradient information of the pyramid fusion image. According to the method, the underwater image is enhanced, so that the contrast and saturation of the enhanced image can be well considered, image details are richer and clearer, and the contrast, information entropy, color information and the like of the whole image can be further improved.

The invention discloses a landslide mass recognition method based on Laplacian pyramid remote sensing image fusion. The method comprises the steps: reconstructing an original remote sensing image through a Laplacian pyramid fusion module according to the extracted local features and global features of the remote sensing image, and generating a fusion image; constructing a deep learning semantic segmentation model through a semantic segmentation network; then respectively marking places where landslide disasters occur and places where landslide disasters do not occur in the fused image through a picture marking tool to obtain a landslide disaster tag map data set; and finally, training a deep learning semantic segmentation model by using the data set, and storing the model by modifying a semantic segmentation network structure and adjusting model parameters until a loss curve of the model reaches fitting and when the precision of recognizing the landslide mass in the remote sensing image meets the requirement. According to the method, an image fusion model based on the Laplacian pyramid is combined, and an effective decision basis can be efficiently and accurately provided for disaster prevention and reduction of landslide disasters.

The invention discloses a multi-exposure image ghosting-free fusion method under global gradient based on patch alignment, which comprises the following steps: reading a reference image, measuring the similarity between the reference image and an LDR image based on a multi-source bidirectional similarity measurement algorithm MBDS, and aligning a motion area in the LDR image by adopting a patch acceleration method; adopting a reconstruction algorithm to obtain an LDR image sequence aligned with the reference image; designing a pixel relative intensity weight formula and a global gradient weight formula; and carrying out weighted averaging on the two weight expressions to obtain a final weight expression, inputting the weight map and the LDR image sequence in the Laplacian pyramid to carry out image fusion, and outputting a fused image. According to the method, the problem of artifacts occurring under dynamic scene fusion is effectively solved, the LDR image is registered based on the reference image, the fusion time is saved, the robustness is higher, then multi-scale decomposition fusion is carried out in the Laplacian pyramid, the fusion effect is better, and the obtained HDR image is rich in detail information and better in visual effect.

The invention discloses a multi-exposure image fusion method and system and a storable medium, and belongs to the technical field of image processing, and the method comprises the steps: inputting an image, calculating a good exposure estimation weight and a chroma weight, and carrying out the convolution of a Gaussian pyramid of a comprehensive weight map and a Laplacian pyramid of the input image, and obtaining an initial fusion pyramid; secondly, carrying out local detail enhancement on the obtained brightest and darkest regions of interest by utilizing a fast local Laplacian filter (FLLF), then taking the maximum value of Laplacian coefficients to obtain a maximum coefficient Laplacian pyramid, updating the coefficients in the regions of interest into an initial fusion pyramid, and obtaining a maximum coefficient Laplacian pyramid; according to the method, under the condition that it is guaranteed that information is not lost in the brightest and darkest areas, proper enhancement can be carried out, the naturalness of light and shadow transition is guaranteed, and image details are clearer.

Systems and methods for providing low dynamic range images from wide dynamic range images. A wide dynamic range image is first converted into a normalized image and is decomposed into a multiple Laplacian images and each of the Laplacian images is passed through one level of the process. Each level of the process has multiple sets of processing layers and produces a transition image. The various transition images form a decomposed Laplacian pyramid of the normalized image and a reconstructed image from the various Laplacian images is the low dynamic range image. Each level of the process is constructed as a neural network whose relevant filters, weights, and biases are determined by training the neural network using manually selected input and output images.