77 results about "Reflectance map" patented technology

The present invention discloses a high-seed super-resolution MIMO (multi input multi output) array imaging method. The method includes the following steps that: step 1, data sampling is performed on an MIMO array; step 2, an MIMO-RMA (range migration algorithm) is utilized to complete three-dimensional reflectivity image reconstruction; step 3, the acceleration of the MIMO-RMA is utilized to calculate a coherence factor; and step 4, the coherence factor corrects a reflectivity three-dimensional image. According to the imaging algorithm of the invention, reflectivity incoherent power is simplified by rearranging echo data, the acceleration of the MIMO-RMA is utilized to calculate the coherence factor, and the calculation speed of the coherence factor is greatly increased in combination with a parallel algorithm; super-resolution performance is balanced through controlling parameters alpha and beta according to a defect that the coherent factor suppresses weak scattering points, and at the same time, the stability of the algorithm is not affected; and the coherence factor is utilized to correct the reconstructed three-dimensional image, so that sidelobe and ground noises can be effectively suppressed, and higher resolution can be achieved.

The invention discloses a fruit distinguishing and locating method with laser scanning and machine vision combined. According to the fruit distinguishing and locating method with laser scanning and machine vision combined, through the characteristics that laser scanning is not influenced by the light condition and can simultaneously obtain multiple pieces of information such as the distance, the reflectivity and the coordinate, fruits are distinguished from branches and leaves in a reflectivity graph generated after scanning through a clear outline formed by the obvious difference of reflectivities. According to the color characteristic difference between a color image and a characteristic point of the reflectivity graph, the mature fruits and the immature fruits are distinguished, wherein the mature fruits and the immature fruit are similar in reflectivity. According to the difference distances of multiple mutually-overlapped fruits in a depth map, the mature target fruit is distinguished from the fruits overlapping with the mature target fruit. According to the fruit distinguishing and locating method with laser scanning and machine vision combined, the multiple pieces of information, obtained after laser scanning, such as the reflectivity, the depth and the coordinate and the color image information of machine vision are combined, the purposes that the mature target fruit is distinguished from the branches and the leaves, the immature fruits, and the fruits overlapping with the mature target fruit and is located are achieved comprehensively. The fruit distinguishing and locating method with laser scanning and machine vision combined is simple and rapid, high in real-time performance and practicability, and applicable to distinction and locating of the fruits on a plant.

The invention discloses an underexposure image recovery method based on deep learning, and belongs to the field of image quality enhancement. The problems of low shooting contrast, low brightness, much noise, high complexity of a mainstream image recovery algorithm and high calculation cost caused by uncomfortable light source in the shooting process are solved. The method mainly comprises the following steps: firstly, the input of a multi-scale decomposition network for shooting images with different illuminance is a low-illuminance image, and a series of feature maps smaller than the original image are obtained through down-sampling convolution; secondly, the reflectivity and illuminance of images with different resolutions are obtained at an up-sampling decoding end through decomposition, on the basis of the decomposition, image brightness enhancement is carried out by using a multi-scale brightness adjustment network oriented to image illumination imbalance, and then reconstructionand recovery are carried out by using an image reconstruction network; finally, the reflectivity image and the enhanced illuminance image are integrated to obtain a final restored illumination image.

A method for ascertaining the position of an edge in or on an object surface region of interest by optical scanning. The reflectivity of the object surface region is evaluated. Light is emitted onto the object surface region under different illumination conditions, in particular different light incidence directions, and in each illumination condition a sequence S1 to Sn of camera images B is recorded. Each camera image B of a sequence S1 to Sn is recorded at another illumination intensity I. Subsequently in each case one reflectivity image R1 to Rn is produced from a plurality of or from all camera images B of a sequence S1 to Sn. Thereafter, a resulting reflectivity image E is produced from a plurality of or from all reflectivity images R1 to Rn by weighted addition, in which resulting reflectivity image E the position of an edge is determined.

The invention belongs to the technical field of digital image processing, and discloses an RGBD image ambient light editing method and system based on spherical harmonic illumination, and the method comprises the steps: carrying out the preprocessing of an inputted RGBD image to obtain a shadow-free reflectivity image, a shadow-free brightness image and a shadow image; decomposing the shadow-freebrightness image into an illumination detail image and a spherical harmonic illumination image; obtaining an adjusted spherical harmonic illumination graph through interactive editing; adjusting the shadow map according to the spherical harmonic illumination map and the adjusted spherical harmonic illumination map to obtain an adjusted shadow map; and according to the shadowless reflectance map, the adjusted spherical harmonic illumination map and the adjusted shadow map, synthesizing to obtain an RGB image edited by the ambient light and outputting the RGB image. The system comprises a preprocessing unit, a bright and dark image decomposition unit, an interactive editing unit, a shadow image adjusting unit and an image synthesis unit. The method is suitable for various conditions with complex ambient light, and can automatically estimate the ambient light in the image.

Method and/or apparatus embodiments for automatically identifying caries, executed at least in part on data processing hardware, can acquire, multiple reflectance images of a tooth, wherein each reflectance image includes at least red, green, and blue intensity values for a region of pixels corresponding to the tooth. Each reflectance image in the plurality of reflectance images is processed by (i) defining a tooth region within the reflectance image; and (ii) identifying caries within the defined tooth region for each reflectance image. The reflectance images are registered and the registered processed images are combined to form a result image that shows the identified caries. In one embodiment, multiple reflectance images of a tooth are taken from substantially the same camera position.

The application, which relates to the technical field of unmanned aerial vehicles, provides a multispectral imaging system and method. The multispectral imaging system comprises a controller, a multi-view camera, an illumination sensor and a digital processor, wherein the multi-view camera, the illumination sensor and the digital processor are electrically connected with the controller. The controller is used for controlling the multi-view camera to shoot a target plot when an unmanned aerial vehicle flies to a set shooting point to obtain a plurality of spectral images; the illumination sensor is used for measuring an illumination intensity value of each preset waveband; and the digital processor is used for performing radiation correction on each spectral image according to the illumination intensity value of each preset waveband to obtain a plurality of reflectivity images, acquiring a target reflectivity image corresponding to a vegetation index from the plurality of reflectivity images according to the vegetation index, and performing index calculation to obtain a vegetation index map. According to the application, the vegetation index map reflecting the plant information in the target plot can be automatically generated without manual participation, so that human resources are saved, and the detection efficiency is improved.

The invention discloses a dark light image adaptive enhancement method based on dense deep learning, and belongs to the field of computer vision. The dense deep convolutional neural network model provided by the invention comprises four sub-network models, so that local adaptive illumination enhancement, noise elimination, distortion repair and semantic repair of the dark image can be realized. The method comprises the following steps: firstly, decomposing a reflectivity diagram and an illumination diagram of the dark light image through a decomposition network; learning noise components with relatively small space in a reflectivity graph obtained through decomposition by combining illumination graph guidance information obtained through an MAMI module and a residual idea, and carrying out noise reduction; then, recovering semantic information of the reflectivity graph through a two-layer dense convolutional neural network and a loss function with a semantic recovery item, and restoring color distortion and the like; introducing the proportion information of the to-be-enhanced illuminance map and the target illuminance map and the gradient information of the to-be-enhanced illuminance map into the illuminance adjustment network, so that the global brightness can be flexibly adjusted and the local brightness can be adaptively enhanced; and finally, synthesizing the enhanced illumination image and the recovered reflectivity image to obtain a target image. According to the method, noise can be removed and distortion and semantics can be restored while the brightness of the image can be adaptively adjusted, so that the enhanced image is excellent in image aesthetic evaluation indexes and is superior to other methods in many computer vision tasks.

The invention relates to a defect detection method and device, a control device and a readable storage medium, and the method comprises the steps: determining a reflectivity graph and a Gaussian rategraph of a to-be-detected target according to an obtained image set of the to-be-detected target based on a photometric stereo method; judging whether the to-be-detected target has suspected defects or not according to the reflectivity graph and the Gaussian rate graph; and if the suspected defect exists in the to-be-detected target, determining the type of the suspected defect according to the binarized image corresponding to the to-be-detected target. According to the defect detection method provided by the invention, after it is determined that the suspected defect exists in the to-be-detected target, the type of the suspected defect can be determined through the binary image of the to-be-detected target, that is, dust and pits in the suspected defect can be determined and distinguished, and misjudgment can be avoided.

The invention provides a human body scene image eigen-decomposition and relighting method and device, and the method comprises the steps: making a human body model through employing Clo3d, carrying out the light tracing through employing blender, and carrying out the rendering, and obtaining a training data set; pre-training a normal prediction network by using a single image; taking the original image and the normal graph obtained through prediction as input of an image eigen-decomposition network, and obtaining a reflectivity graph, a shadow graph and spherical harmonic illumination parameters through prediction; re-illuminating the original image by using the new spherical harmonic illumination parameter; and finally, decomposing a single picture into a reflectivity picture and a shadow picture, so the application of picture relighting, 3D model synthesis, model surface texture reconstruction and the like can be achieved. The shadow of the input picture can be removed in the field of three-dimensional reconstruction, and the surface texture quality of a three-dimensional reconstruction model is improved.

The invention discloses a long-distance photon counting three-dimensional laser radar imaging system and method, and the system obtains the flight time data of a target photon through employing a time-dependent photon counting technology in combination with an avalanche photodiode working in a Geiger mode, and carries out the analysis of original data through a special algorithm and software, therefore, the depth and reflectivity images of the long-distance target can be accurately reconstructed. According to the system, the problems that a traditional photon counting laser radar system is short in action distance and low in imaging speed are solved, and the application range of the photon counting laser radar system is effectively widened.

The inventive system and method presented herein combines and integrates information to provide a more effective single rating of vegetation health and biomass. As a result the present invention provides for an objective biomass rating which produces a single numerical measure of vegetation biomass and health that can be used for quantitative comparative assessment of vegetation in research trials and in plant production. The inventive system and method further provides for remote-sensing of vegetation health and biomass based on combined assessments of morphological data (extracted from sources such as high-resolution elevation raster-maps or point-clouds produced by photogrammetric or LiDAR sensors) and radiometric data (such as spectral reflectance maps generated from multi- or hyper-spectral imaging systems).

The invention relates to a cotton vegetation index remote sensing detection method based on a Sentinel-2 satellite. The cotton vegetation index remote sensing detection method comprises the followingsteps of acquiring Sentinel-2 satellite remote sensing image data of cotton in different growth stages in a research area; performing radiation correction and atmospheric correction of the remote sensing image data; determining a reflectivity image of the remote sensing image; through distributing Sentinel-2 multispectral satellite data in a plurality of wave bands from visible light to near infrared, obtaining multiple vegetation indices. The method is advantaged in that the Sentinel-2 satellite data has high spatial-temporal resolution and a specific red edge parameter waveband, ideal data sources can be provided for monitoring of different growth stages of cotton, vegetation indexes of different growth stages of cotton can be extracted more accurately and rapidly, the operation processis simple and flexible, detection efficiency is effectively improved, economic cost is low, powerful tool support is provided for timely analysis of physiological parameters of cotton, and the methodhas high application and popularization values.

The invention discloses an interactive relighting editing method based on an RGB-D image. The method comprises the following steps of: preprocessing an input image to obtain a shadow-free reflectivity image, a bright-dark image and a shadow image; segmenting an original RGB image by using a MaskRCNN algorithm to obtain a segmentation mask of an object in a scene, and then obtaining a corresponding local depth map, a shadow-free reflectivity map and a shadow-free bright-dark map; decomposing the shadow-free bright-dark image into a global illumination detail image, a local illumination detail image and a spherical harmonic illumination image; sampling and visualizing the global spherical harmonic illumination image and the local spherical harmonic illumination image, and obtaining an adjusted global spherical harmonic illumination image and an adjusted local spherical harmonic illumination image through interactive editing; and synthesizing the global and local shadow-free reflectivity images, the adjusted spherical harmonic illumination image and the shadow image to obtain global and local RGB relighting images subjected to ambient light editing, and outputting the global and local RGB relighting images. The indoor complex scene image enhancement under a low-light condition can be effectively, conveniently and intuitively realized.

An apparatus for acquiring intraoral images of a subject has an OCT imaging apparatus having an OCT light source, a scanner that conveys OCT light toward the subject and returned from the subject, and an interferometer having a reference arm and a sample arm. A reflectance imaging apparatus has a visible light source directed toward the subject and an image sensor that forms a reflectance image from returned light. Processing and control logic are configured to process and combine the returned reflectance image to the OCT measured data. A display shows the combined reflectance image and OCT measured data.

The invention discloses an automatic extraction method and system for a sub-pixel level water body from a medium-low resolution remote sensing image, and the method comprises the steps: carrying out the preprocessing on the given medium-low resolution remote sensing image, and generating a surface reflectance image; extracting a normalized difference water body index from the surface reflectance image and stretching the normalized difference water body index to construct a stretched water body index image histogram; describing a stretched water body index image histogram by using a Gaussian mixture model, and carrying out estimating to obtain water body and land parameters; determining a water and land separation threshold value according to the water body and land parameters; extracting awater and land mixed pixel according to the water body and land parameters and the water and land separation threshold; and carrying out water abundance calculation on the land-water mixed pixel to obtain a sub-pixel-level water extraction result. According to the method, the water-land separation threshold value can be rapidly and accurately extracted and is not influenced by human factors; andmeanwhile, the problem of missing extraction or over-extraction of small water bodies on the medium-low resolution remote sensing image is solved effectively.

The invention discloses a method for estimating the sea surface green tide coverage area under MODIS image cloud, and the method comprises the following steps: carrying out the preprocessing of remotesensing image data, obtaining a water body remote sensing reflectivity image and a DVI image map of a research sea area, extracting green tide information and surrounding cloud information covered bygreen tide, determining a cloud pixel with green tide distribution around by utilizing the rectangular frame, assuming that the cloud pixel is a central pixel and the same rectangular frame with green tide at the periphery is an adjacent pixel, calculating a cloud coverage area in the central pixel and a green tide coverage area in each adjacent pixel, calculating the mean value of the green tidecoverage ratios in the adjacent pixels, taking the mean value as a central pixel value, and taking the product of the cloud coverage area in the central pixel and the central pixel value as the coverage area of the green tide on the sea surface under the cloud. According to the method disclosed by the invention, the green tide coverage area of the sea surface under the MODIS image cloud can be better estimated, the influence of the cloud on the optical image on green tide area estimation is reduced, and technical support can be provided for quantitative analysis and research of green tide.

The invention belongs to the technical field of remote sensing identification of mineral resources, and discloses a traffic difficulty area sedimentary carbonate lead-zinc ore survey method and system, equipment and application. The method comprises the steps: acquiring a high-resolution image and a remote sensing image, wherein the remote sensing image is a multispectral satellite remote sensingimage; performing image processing on the high-resolution image, and determining a limestone distribution area; interpreting the processed high-resolution image according to the limestone distributionarea to obtain an interpreted high-resolution image; performing radiation correction and atmospheric correction processing on the remote sensing image to obtain a remote sensing image of which the pixel value is reflectivity; performing matched filtering operation on the remote sensing image reflectivity image in the interpreted limestone distribution area to obtain a matched filtering grayscaleimage; performing threshold division on the grayscale image, and dividing a high-value part representing the position of a lead-zinc ore deposit; according to the method, the secondary oxidized carbonatite type lead-zinc deposit can be efficiently, quickly and accurately identified.

The invention discloses a standard reflection-type supercontinuum spectrum imaging instrument, and the instrument consists of an embedded main controller, a terahertz scanning imaging subsystem, a multispectral imaging subsystem, a sample room, and a wireless local area network transceiver. The instrument provides a supercontinuum spectrum dodging laser source and a terahertz single-beam focusingscanning illumination method, and combines standard black and white boards for correction. Because the geometric shapes of feather and skin samples of rare animals cannot cover the whole image, the contour of the whole sample can be easily obtained through black base correction, and a standard reflectivity image of the sample can be obtained through white base correction. An active illumination imaging mode eliminates the impact from the ambient light, and the visible light, infrared and terahertz multi-band standard reflectivity images correlated with the tissue characteristics of the sampleare obtained. The instrument brings convenience to a custom import and export detection sanitary authority for the tracking, discrimination and protection of important biological resources.

The invention provides a polarization denoising method for single photon counting imaging, which comprises the following steps of: calculating the total intensity detected by detectors in two directions through a stokes vector and a Mueller matrix, and defining a new reflectivity index alpha N; constructing a single photon counting imaging polarization denoising model by calculating a negative logarithm likelihood function of the reflectivity alpha and the linear depolarization parameter beta and adding a non-local prior sparse representation constraint; and by solving the optimization problemthrough an alternating minimization method, obtaining and outputting a reflectivity and depolarization parameter image. According to the invention, in a high background noise environment, the polarization difference between signal light and background light is utilized, so the signal-to-noise ratio is improved fundamentally, and meanwhile, non-local prior sparse representation constraints are combined, so that not only can a high-quality reflectivity image reflecting target features be obtained, but also a linear depolarization parameter image of a target can be obtained, and polarization information of the target is obtained.

The invention discloses a rapid detection method and system for multiple defects on the surface of an aluminum valve plate, a medium and a terminal, and relates to the technical field of visual optical design. Aiming at a to-be-detected position on the surface of the aluminum valve plate, the camera and the light sources are controlled by the manipulator to reach a specified position, the same position on the surface of the aluminum valve plate is sequentially irradiated by the eight light sources at four angles of the same plane, and the camera sequentially photographs along with the irradiation of the different light sources to obtain images of the light sources at the corresponding angles; real-time integration processing is performed on the image corresponding to each light source angle, namely the image is calculated by using a photometric stereoscopic vision theory to obtain a corresponding illumination reflectivity image and a Gaussian convolution graph of the image; and a preprocessing area of the image is obtained through the albedo image and the Gaussian convolution image, polar coordinate conversion is carried out on the preprocessing area, and filtering processing is carried out after polar coordinate conversion. The problems that some qualified and slight scratches are not obviously compared with unqualified scratches on the image, and unqualified products cannot be recognized are solved.