85 results about "Image Quantification" patented technology

A computer-implemented method for quantifying an object in a larger structure using a reconstructed image obtained from an imaging apparatus includes identifying an object boundary in a region of interest of the image and a region of exclusion, and determining a mean value of intensity in a neighborhood of the object to estimate a background value, excluding the region of exclusion. The method further includes determining an interior region, a transition region, and a background region of the object using the object boundary and a transfer function of the imaging apparatus, and determining and displaying a measure of object quantification utilizing intensities of at least one of the determined interior region or the transition region a method for quantifying an object in a larger structure using a reconstructed image obtained from an imaging apparatus.

A method and apparatus are disclosed for testing the accuracy of digital test images by comparing the value of each test image pixel with the average value of a corresponding reference image pixel array. The local image quantification verification test program accepts absolute differences between corresponding pixel and array values that are less than a specified difference maximum. The user may specify a difference maximum and a maximum array size, and may restrict the test image comparison to specified regions of the test image with specific difference tolerances for each region. This allows the user to focus the comparison on certain regions as more important than others and to focus a region on a particular object or group of objects.

The invention relates to an image retrieval method and apparatus. The method comprises the steps of: obtaining a to-be-retrieved image from a user terminal; converting the to-be-retrieved image to HSV space from RGB space; quantizing the converted to-be-retrieved image into N-dimensional color features; sorting the feature values of the N-dimensional color features in a descending order, and selecting the first M-dimensional color features as main color of the to-be-retrieved image; determining a cluster index table name of the to-be-retrieved image according to an index value corresponding to the main color; querying whether a cluster index table with the same name as the cluster index table of the to-be-retrieved image exists in the established cluster index table according to a name of the cluster index table; if existence, obtaining an image index of the corresponding image; querying main color percentage of the corresponding image according to the obtained image index; calculating the image similarity according to the main color percentage; and returning an image matched with the to-be-retrieved image to the user terminal according to the similarity. The image retrieval method and apparatus disclosed by the invention remarkably improve the image retrieval efficiency and accuracy.

The invention discloses an image color retrieval method and an image color retrieval system and relates to the field of image processing. The method comprises the steps of performing quantification processing to HSV (Hue, Saturation and Value) color spaces of pixels in an image to obtain quantified color values within a range of 0 to N, wherein non-uniform fuzzy quantification is performed to tone values; statistically collecting a histogram of the quantified color values of the pixel of the image and performing normalization processing by using the sum of the pixels of the image to obtain a color characteristic vector of the image; comparing the color characteristic vector of the image to the color characteristic vectors of other images to determine the similarity of the images. The method and the system disclosed by the invention have the advantages that since images related to the inquired image color are extracted from a database by taking the color characteristic vector after image quantification as a relevant judgment basis, the image retrieval is enabled to be more intelligent and the extraction result of color-related images becomes more accurate; since the images in the database are processed in advance by adopting an offline module, the execution speed is effectively improved.

The present invention discloses an urban high-resolution remote sensing image shadow detection and segmentation method. The method includes the following steps that: image quantification is performedon urban multi-band remote sensing images, shadow detection and compensation are performed on the quantized remote sensing images on the basis of chi-square transformation, so that shadow compensatedimages can be obtained; and a multi-scale J-image sequence is calculated for the obtained shadow compensated images, and multi-scale segmentation and region combination are performed, so that final remote sensing image segmentation results can be obtained. With the method of the invention adopted, weak edges and false edges caused by shadows can be can effectively dealt with. The segmentation accuracy of the method is significantly improved; and the method has high reliability.

A video encoder identifies scene changes during the encoding of video sequences based on the distribution of macroblock types within each image. Once an image is identified as a scene change, control information is sent to a quantizer to adjust the quantization of the image to improve image quality. The image quantization is improved without changing the frame or picture type of the image within its group of pictures. To improve the image quality, the quantizer can directly or indirectly change the quantization parameter used to quantize the DCT coefficients for the image.

The present invention relates to the embedding and extracting method of wavelet chaotic semi-fragile digital water print. Original image is 2D two-stage wavelet analyzed to obtain {LL2, HL2, LH2, HH2, HL1, LH1, HH1}, where, the pixel grey of LL2 sublayer is mapped to form chaotic initial value and chaotically iterated for several times to form water print image. Then, based on the quantized medium frequency coefficients { HL2, LH2, HH2}, and by means of the extreme sensitivity of chaos to the initial value and the excellent temporal frequency characteristic of wavelet, the distortion to water print image is precisely located. Via flexible regulation of quantized parameters, the robustness against baleful attack is obtained. The method of the present invention has stable performance, fastrealizing speed, strong robustness to baleful distortion and excellent distortion locating and proving capacity.

The invention discloses an image bit depth enhancement method based on a deep learning network, and the method comprises the following steps: carrying out the preprocessing of a high-bit lossless-image-quality image, and carrying out the quantification of the high-bit lossless-image-quality image into a low-bit image; designing a convolutional neural network based on deep learning, taking the quantized low-bit image as input, and taking perception loss between an output result and the original high-bit image as a loss function; training parameters of each convolution layer and each batch normalization layer in the model through an optimizer gradient descent loss function, and storing the corresponding model and parameters of all layers in the model after the descent amplitude of the modelloss function does not exceed a threshold value; and after the high-bit image is quantized into low bits, recovering the high-bit image through the stored convolutional neural network. According to the method, a convolutional neural network framework in deep learning is utilized, so that a high-quality high-bit image can be accurately recovered.

The invention relates to an automatic classification method for breast tumor ultrasound images, and belongs to the technical field of applications of artificial intelligence. According to the invention, quantitative features of the breast tumor ultrasound images are tightly combined, histogram features, color features, profile features, edge features and echo features in the breast tumor ultrasound images are quantified, a decision-making tree, a Naive Bayesian method and a random forest method are selected to serve as machine learning methods, and a decision-making tree, Naive Bayesian and random forest weighted fusion multi-dimensional classification method is proposed to perform recognition on the breast images. The automatic classification method can effectively improve the automatic classification accuracy of the breast tumor ultrasound images.

The invention relates to an automatic evaluation method of image quantitation of a medical magnetic resonance model body. The automatic evaluation method comprises the following steps: by adopting image skewing correction, performing two-dimensional rotation to images, and interpolating images by adopting a bicubic interpolation method to ensure that the images are in normotopia; calculating FWHM of each layer thickness area by using a method of calculating the average of a plurliayt of profile lines, so as not to be influenced by subjective factors; increasing image uniformity profile curves, so as to enable the distribution uniformity of images to be clear; extracting the profile of step edges of an ROI area in high-contrast resolution detection by using a watershed method, accurately positioning the position areas of a plurality of line pair sets, so as to accurately draw profile lines; accurately positioning the maximum circular spot in each set of low-contrast resolution detection area by using an image segmentation algorithm of a horizontal set, and automatically and accurately positioning the centroid of all the ROI areas of geometric distortion detection layer by using a centroid method, to guarantee the geometric distortion rate and aspect ratio accuracy. The automatic evaluation method does not rely on individual experiences, the detection speed is fast, and the detection efficiency and the detection accuracy can be greatly promoted.

The present invention discloses an underwater sound image compressing method based on discrete wavelet conversion under high error bit ratio, including the following steps: 1) to statistical analysis the child band energy of image wavelet parameter, and obtain wavelet parameter child band energy statistical table by implementing three-layer decomposing to the underwater sound image through dual orthogonal CDF9/7 wavelet group; 2) image quantification; 3) encoding and decoding, to implement compress encoding to the quantification parameter after the three-layer decomposing. The encoding process includes the following steps: (1) to implement definite length encoding to the parameter of the child band LL3, HL3, LH3, HL2 and LH2; (2) to divide the child band LH1 into sixteen subblocks of sixteen by sixteen, each subblock uses the front 128 zero journey encoding mates, the journey number and the non zero value is defined by 2bits and 3bits separately; (3) to abnegate the encoding to the child band of HH3, HH2, HH1 and HL1. The method of the present invention counterpleas high error bit ratio, the image quality after compressing is very good.

The invention discloses an image quantitative analysis method. The method comprises the following steps of obtaining an original image sample, wherein the original image sample includes an original training sample and an original test sample; performing smooth fuzzy processing, normalizing image block data in the original training sample and the original test sample to reduce the resolution of theimage for obtaining a fuzzy training sample and a fuzzy test sample, and enabling the fuzzy test sample to have the same data features as the fuzzy training sample; and training a super-resolution reconstruction model, processing the fuzzy test sample, and constructing a residual error spectrum, wherein a high-gray value response region in the residual error spectrum is namely an abnormal featureregion.

The invention discloses an image bit enhancement method based on multilayer features of a series neural network. The image bit enhancement method comprises the following steps: constructing a trainingset, quantizing a high-bit image of the training set into a low-bit image, solving a difference between the high-bit image and the low-bit image according to pixels to obtain a residual image, and performing zero filling on the low-bit image to obtain a zero-filled high-bit image; removing random variables in the VAE network, directly inputting a feature map generated by an encoder into a decoder, and establishing a deep learning network model on the basis of the feature map; adding a plurality of series jump connections into the network model, and transmitting each layer of feature map to all subsequent layers; inputting the zero-filling high-bit image into a deep learning network model to generate a residual image, and training a network by using an Adam optimizer; and quantizing the high-bit image of the test set into a low-bit image, inputting the zero-filling high-bit image into the network loaded with the training model parameters to generate a residual image, and adding the residual image and the low-bit image according to pixels to obtain a reconstructed high-bit image.

The invention relates to an image quantization method, computer equipment and a storage medium. The method comprises the following steps: segmenting a medical image of a to-be-detected object to obtain a first class segmented image and a second class segmented image, wherein the first class segmented image comprises at least one first region of interest, and the second type segmented image comprises second regions of interest corresponding to the first regions of interest; determining feature data corresponding to each second region of interest according to the at least one first region of interest and the second region of interest corresponding to each first region of interest, wherein the feature data is used for representing the distribution condition of each second region of interest;and determining a quantized value corresponding to the medical image according to the feature data corresponding to each second region of interest and the clinical feature data of the to-be-detected object, wherein the clinical feature data is data obtained after clinical detection is conducted on the to-be-detected object, and the quantized value is used for indicating the severity degree of pulmonary infection on the medical image. The method can improve the detection precision.

The invention discloses a real-time multi-frame bit enhancement method based on content and continuity guidance. The method comprises the steps of quantizing a high-bit image into a low-bit image, carrying out low-bit zero padding of the low-bit image, and obtaining a zero-filled high-bit image as a training set of a network; removing a motion compensation module and a time sequence symmetric sub-network in the SS-VBDE network, and connecting the feature maps, meeting the symmetric positions in space, of a convolution layer and a deconvolution layer to realize spatial symmetric jump connection; performing low-bit zero filling on the low-bit image of the training set to obtain a zero-filled high-bit image as network input; combining the image content loss and the inter-frame continuity lossbetween the multi-frame image sequence generated by the improved network and the real image sequence to serve as a loss function, and training improved network model parameters through an Adam optimizer; and performing low-bit zero filling on the low-bit image of a test set to obtain a zero-filling high-bit image, inputting a high-bit image sequence into the improved SS-VBDE network after the network model parameters are loaded, and outputting a processed high-bit image sequence.

The invention provides an extraction method of a histogram texture descriptor in a muti-contrast mode, comprising the following steps: carrying out image quantization; separating the quantized image into a positive matrix, a negative matrix and an equivalent matrix; and carrying out histogram calculation according to the three matrixes to obtain the histogram descriptor in a local mode of the image. The technical proposal provided by the invention can achieve efficient and simple extraction of textural features and well express the texture of an SAR image.

The invention relates to a nuclear medicine heart image quantification technology, in particular to to a technological method for multi-pinhole SPECT or SPECT/CT dynamic image quantitative reestablishment and myocardial blood flow absolute quantification measurement and application of the technological method in myocardial blood flow state evaluation. The technological method includes the specific steps of nuclide physical decay correction, in-scanning patient movement correction, scattering correction, geometric warping correction, data truncation compensation, tissue attenuation correction, noise removal, pixel value conversion, myocardial blood flow quantitative calculation and blood flow state evaluation. With the technological means, a quantitative multi-pinhole SPECT dynamic heart image can be generated, myocardial blood flow absolute quantification calculation can be carried out through the quantitative multi-pinhole dynamic image, and thus quantitative measurement of the myocardial blood flow is achieved; meanwhile, a blood flow state diagram is established with the three indexes of resting blood flow, load blood flow and blood flow reserve and is put into the practical application of evaluating the myocardial blood flow state, the purpose of carrying out quantitative measurement of the myocardial blood flow with multi-pinhole SPECT and SPECT/CT dynamic development is achieved, and the technological method can be applied to evaluation of the myocardial blood flow state.

Computerized image guided control method uses image quantification to dynamically monitor and control the cell generation process. The method acquires images during the cell generation steps and uses computer image analysis to automatically quantify the results of the cell generation steps. In one embodiment, the quantified results from the image analysis are used to determine the readiness of the cell generation step. In another embodiment, a remedial recovery sub-step for the cell generation step is also guided by the imaging results. In yet another embodiment, when a cell generation step is determined to have failed, the cell generation process is early terminated based on image guided decision.