126 results about "Homomorphic filtering" patented technology

The subject invention pertains to a method of image intensity correction. The subject invention can utilize extrapolation for image intensity correction. The use of extrapolation can reduce the artifacts during intensity correction as compared to traditional methods of intensity correction. The extrapolation can be combined with, for example, homomorphic filtering methods, parametric estimation techniques, wavelet based method, and/or Gaussian smooth method, in order to reduce the artifacts generated by these methods and improve the quality of correction. The implementation of image extrapolation in accordance with a specific embodiment can utilize closest point method. The subject method can also use adaptive smoothing for image intensity correction. In an embodiment, the use of gradient weighted smoothing method can reduce, or eliminate, over-smoothing of bright spot regions. In a specific embodiment, the subject method can utilize gradient weighted partial differential equation (PDE) smoothing.

The subject invention pertains to a method of image intensity correction. The subject invention can utilize extrapolation for image intensity correction. The use of extrapolation can reduce the artifacts during intensity correction as compared to traditional methods of intensity correction. The extrapolation can be combined with, for example, homomorphic filtering methods, parametric estimation techniques, wavelet based method, and/or Gaussian smooth method, in order to reduce the artifacts generated by these methods and improve the quality of correction. The implementation of image extrapolation in accordance with a specific embodiment can utilize closest point method. The subject method can also use adaptive smoothing for image intensity correction. In an embodiment, the use of gradient weighted smoothing method can reduce, or eliminate, over-smoothing of bright spot regions. In a specific embodiment, the subject method can utilize gradient weighted partial differential equation (PDE) smoothing.

The invention discloses a vehicle type recognition method based on vehicle face features. The method comprises the following steps of: (1) collecting vehicle images through a monitoring camera, preprocessing the images and detecting and segmenting vehicle face images capable of characterizing vehicle types, wherein the step (1) comprises the following specific procedures of: (1-1) collecting the vehicle images of various types of vehicles by using the monitoring camera, (1-2) preprocessing the images and enhancing the quality of the images through homomorphic filtering, and (1-3) segmenting vehicle face regions capable of characterizing vehicle type features by adopting vehicle face region detection and segmentation based on license plate position information; (2) carrying out Curvelet wavelet transform on the vehicle images so as to extract a vehicle face feature matrix capable of characterizing the vehicle type features; and (3) classifying the Curvelet wavelet feature vectors of the extracted vehicle face images by using a support vector machine classifier so as to recognize the vehicle type. The method provided by the invention can be used for providing more accurate vehicle and vehicle type information for traffic monitoring and is very important for traffic safety and the real-time extraction of traffic information.

The invention relates to a method for detecting citrus fruits on a tree. The method comprises the steps that the local and global information entropies are combined to analyze the illumination distribution characteristic of an image, and a local block homomorphic filtering algorithm is used to process illumination unevenness of the image; an adaptive enhanced RG color difference method is used toimprove the contrast between a foreground target and the background; threshold segmentation and a mathematical morphology method are used to realize the binarization of an RG chromatic aberration diagram; for the problem of multi-fruit adhesion or mutual occlusion in an acquired binary image, marker-controlled watershed transformation is used to realize adhesion region segmentation; a convex hulltheory is used to solve the problem of region under-segmentation caused by occlusion to acquire an independent candidate citrus fruit region; the histogram description feature of the candidate citrusfruit region is extracted; and citruses are identified by a histogram cross-core support vector machine to complete fruit location. According to the technical scheme provided by the invention, the accuracy and reliability of the detection of the citrus fruits on the tree are improved.

The invention discloses a power transmission line wire defect detection method based on machine vision, and the method comprises the steps: obtaining field image data, and making a training data set;constructing and training an instance segmentation network to obtain a prediction model; deducing an input picture by the model to obtain a rectangular region image and a binarized mask image of the wire; adopting a skeleton algorithm to extract a wire skeleton, calculating the average width of the wire, and reconstructing a binarized mask image; adopting a homomorphic filtering algorithm to eliminate the influence of uneven illumination of the rectangular area image, and combining a reconstructed binarization mask image to extract a segmented wire area image; generating a large number of rectangular frames on the wire area for screening; making a classification training data set, constructing and training a shallow classification network, and obtaining a classification prediction model; inputting the lead section area picture into the classification prediction model, and counting the defect type and defect proportion of the lead section state. The method can accurately segment the wire, detect the state of the wire in a segmented manner, and judge the defect type and defect degree of the wire.

The invention discloses a nighttime double fruit overlap tomato identification method based on overlap edge detection under active illumination. The method comprises the following steps of collecting a nighttime double fruit overlap tomato image under active illumination; using an OTSU method to perform image segmentation based on R-G color difference; performing image enhancement by using a homomorphic filtering method for the grayscale of the double fruit overlap tomato region extracted after image segmentation; extracting the edge, the contour and the overlapping edge of a spot area of the double fruit overlap tomato region after the image enhancement; judging the anteroposterior position relationship between the front unshielded tomato and the later-blocked tomato in the double fruit overlap tomato using a distance nearest method; and circularly fitting the overlapping edge and the front unshielded tomato contour to achieve recognition of the front unshielded tomato in the double overlap tomatoes. The method can realize the recognition of the front unshielded tomato in the double fruit overlap tomatoes at night, and has good recognition effect on overlap tomatoes with different overlapping rates, especially for double fruit overlap tomatoes with the large overlap rate.

The invention discloses an illumination unevenness image enhancement method and system. The method comprises the steps of S1, converting an illumination unevenness image to an HSV color space from an RGB color space, and extracting a brightness component image; S2, performing spatial multi-size homomorphic filtering on the brightness component image to obtain a primary filtered image; S3, performing local entropy-based steerable filtering on the primary filtered image to obtain a final filtered image; and S4, combining the final filtered image as an enhanced brightness component image with an original hue component image and an original saturation component image, and converting a combined image to the RGB color space from the HSV color space, thereby obtaining an enhanced image. According to the method and the system, the illumination unevenness image can be enhanced, so that the illumination is even, the details are rich, and the color is natural.

The invention provides a homomorphic filtering based image processing method and system. The method includes the steps of acquiring an original code map; subjecting the original code map to homomorphic filtering to obtain a code map subjected to homomorphic filtering; based on a preset partitioning algorithm, partitioning the code map subjected to homomorphic filtering into a plurality of code map sub-units; subjecting each code map sub-unit to binarization by an OTSU algorithm to obtain a binary code map sub-unit; combining the binary code map sub-units to generate a final binary code map. The homomorphic filtering based image processing method and system has the advantages that before binarization of the code map, homomorphic filtering is performed first, the images are effectively enhanced by homomorphic filtering, image quality decline caused by insufficiency of light is effectively avoided, the advantages of the OTSU algorithm are given to maximum play, a foreground and a background are precisely distinguished, no effective information is lost from the binary images, image binarization quality is improved, and the following image processing is simplified.

The invention relates to the technical field of vehicle license plate recognition, and discloses a framing method for a license plate in an RGB format. The framing method includes the following steps of converting a license plate image from RGB space to HSV space, separating an HSV component of the image from the image, conducting homomorphic filtering on an luminance component V, converting the image HSV component from the HSV space to an RGB component in the RGB space, determining a luminance threshold value of the RGB component, scanning the license plate image in a line direction and a row direction, recording pixel points of colors dropping into a luminance threshold value range, determining lines and rows where a license plate range is located according to pixel points in each line and pixel points in each row, and displaying pixel points of the lines and rows where the license plate range is located. According to the framing method for the license plate in the RGB format, preprocessing is conducted on a vehicle image with a poor light condition, excessively bright or excessively dark or shaded influence caused by uneven illumination, polarized light, side light, highlight and the like on the vehicle image can be eliminated, and the accuracy rate of license plate location is improved.

The invention discloses an image enhancement algorithm based on wavelet transform and fusing high and low frequency information. The image enhancement algorithm is mainly used for solving the problemof uneven illumination in the fields of medicine, photography and security and protection, comprehensively utilizes an image enhancement method and sequentially comprises the following steps: S1, inputting a color picture into a computer, converting the color picture into a grey-scale map, carrying out two-layer wavelet decomposition of a two-dimensional signal on the grey-scale picture, and converting the image into a frequency domain from a spatial domain; s2, extracting a low-frequency component of the image, and improving high-frequency response in the low-frequency component by adopting homomorphic filtering; s3, converting the filtered low-frequency image into a spatial domain for gray scale conversion to enhance the contrast and brightness of the image; and S4, reconstructing an image according to the enhanced low-frequency component and high-frequency component coefficients, performing wavelet transform-based threshold denoising on the reconstructed image, and finally obtainingan enhanced image. By means of the technical scheme, the low-frequency information, namely the overall brightness of the image, of the image is effectively enhanced, and meanwhile noise interferencein the high-frequency information is restrained.

The invention discloses a detection method for surface defects of an engine cam shaft part. The detection method comprises a detecting algorithm for ring surface defects of upper and lower end surfaces, an algorithm for detecting whether a thread exists at the bottom or not, and an algorithm for detecting whether scrap iron is remained in a bottom hole or not, wherein the detecting algorithm for the ring surface defects of the upper and lower end surfaces adopts a possibility Hough conversion method combining with gradient information to carry out circle center positioning on the upper and lower end surfaces; central circular holes are removed; then images of the ring surface of the upper and lower end surfaces are pre-treated by adopting homomorphic filtering and image noises are removed by using opening operation; defect regions are clustered and defect types are judged. According to the detection method, a coaxial light source, an annular light source, a point light source and an area-array camera are used for carrying out polishing and image acquisition on the end surface of the part and the bottom hole. A dividing device and a photoelectric switch are used for controlling an intermittent rotary mechanical structure of a circular disc, so that the automatic detection, with the resolution rate better than 0.1mm and the single piece detection time less than three seconds, on surface cracks, scratches and spots on the surface of a workpiece, and on whether the scrap iron is remained in the thread of the bottom hole of the part and in the hole or not can be realized.

The embodiment of the invention discloses an image enhancement method and device. The method comprises the steps that brightness component data of images and component data in the color space corresponding to the brightness component data except the brightness component data are extracted; homomorphic filtering is conducted on the brightness component data; contrast-limited adaptive histogram equalization (CLAHE) is conducted on the brightness component data obtained after homomorphic filtering; guide filtering and noise reduction are conducted on the brightness component data obtained after CLAHE to obtain filtered brightness component data; according to the filtered brightness component data and the component data in the color space corresponding to the brightness component data except the brightness component data, enhanced images are synthesized, the dynamic range of the images is effectively widened based on image detail keeping and image noise control, the image contrast is enhanced, meanwhile cloud and mist removal can be effectively achieved, and an image display effect is improved.

The invention discloses a self-adapting enhancement method of an underwater degraded image in the technical field of image processing. The method comprises the steps that a single underwater color image is collected, logarithm transformation is firstly carried out on an input image, then local contrast enhancement processing is carried out on the image obtained after logarithm transformation based on the human eye visual characteristic, and meanwhile illumination estimation is carried out on the image obtained after logarithm transformation through Gaussian homomorphic filtering with a weight; the image obtained after illumination estimation is subtracted by the image obtained after local contrast enhancement; finally, on the basis of analyzing all channel histograms of a deblurred image, color correction and restoration are carried out on the image obtained after multi-scale Retinex enhancement through a color gradation quantization algorithm of the histograms. Image burring caused by the water absorption and particle matter scattering function can be removed well, and the interference of water colors on the color of the image can be removed well.

The invention provides a night complex vehicle detection method. The method comprises: first extracting vehicle lamps by using the homomorphic filtering technology in a frequency domain and the orientation fuzzy technology in a time domain; then performing pairing on the vehicle lamps by using the statistics information of the successfully paired vehicle lamps and introducing a vehicle lamp pairing feedback correction mechanism herein so as to enable a pairing result to be more accurate; and at last tracking vehicles and introducing a vehicle tracking feedback correction mechanism so as to ensure a single corresponding locus for a single vehicle. The shielding problems are respectively handled by using the vehicle lamp pairing feedback correction mechanism for comparison and determination. By the application provided by the invention, higher-accuracy vehicle detection and tracking can be realized, and an overall tracking system has higher precision.

The invention discloses an underwater image enhancement processing method which combines a homomorphic filtering method with a biorthogonal wavelet threshold filtering method. The underwater image enhancement processing method mainly includes the following steps: A), performing logarithm operation on an original image acquired underwater to acquire Z (x,y); B), performing discrete Fourier transformation on Z (x,y) to acquire Z (u,v); C), performing homomorphic filtering processing on Z (u,v) to acquire S (u,v); D), performing Fourier inversion on S (u,v) to acquire s (x,y); E),performing exponent operation on s (x,y) to acquire a processed image g (x,y); F), applying the biorthogonal wavelet threshold filtering method to process the image. By the underwater image enhancement processing method, noise of the image can be removed, contrast ratio of the image is increased, illumination uniformity of a processed underwater image is improved, both balanced filtering and median filtering of a peak signal to noise ratio (PSNR) histogram are improved, underwater image quality is improved, and the underwater image enhancement processing method has certain practical value in improving underwater image quality.

The invention puts forward a hot-line work robot bolt identification method based on random Hough transform and a SVM (Support Vector Machine). The hot-line work robot bolt identification method comprises the following steps that using a binocular camera to collect an operation scene binocular image including bolts; according to the image, judging whether haze is in the presence or not, if the haze is in the presence, adopting a multiscale Retinex method with color restoration to carry out enhancement and restoration on the image, and then, extracting image edges; if the haze is not in the presence, further judging whether a highlight phenomenon is in the presence in the image or not, if the highlight phenomenon is in the presence, adopting a homomorphic filtering method to enhance the image, then, extracting the image edges, and if the highlight phenomenon is not in the presence, directly extracting the image edges; adopting a method based on random Hough transform to carry out ellipse fitting on image edge information; and if a plurality of ellipses are fit, adopting a target identification method based on the SVM to carry out classification identification on a plurality of ellipse fitting results. By use of the method, bolts can be quickly and accurately identified under a complex background.

The invention relates to a fault diagnosis method and system of a planetary gear transmission system, belonging to the fields of a fault diagnosis technique and a signal processing and analyzing technique. The fault diagnosis method comprises the following main fault diagnosis steps: (1) processing acquired vibration signals of the planetary gear transmission system by utilizing a homomorphic filtering technical method, and separating signals comprising fault features and external excitation background noise signals to reduce the complexity of data; (2) carrying out frequency spectrum reconstruction on the signals comprising the fault features after homomorphic filtering processing by adopting complex cepstrum transform, and extracting the fault feature information in the vibration signals of the planetary gear transmission system; and (3) diagnosing the running state of the planetary gear transmission system to be the normal state or the fault state by combining a vibration mechanism of the planetary gear transmission system, utilizing a classifier of a support vector machine (SVM) and taking the time domain statistics at different states and peak value energy after frequency spectrum reconstruction as fault feature vectors.

The invention belongs to the technical field of visual navigation, and discloses a stereoscopic visual navigation method based on green crop feature extraction invariance, which is used for extractinglocal features of green crops, and obtaining the change condition of the number of image feature points during angle change by using Gaussian homomorphic filtering and histogram equalization to process an image; establishing a crop elevation map by using the obtained feature points; extracting the edge contour of the target object in the image, and separating the green crop from the crop background; extracting navigation parameters from the confidence density map fused by the elevation map and the edge map, and adding navigation angle and side edge distance constraint conditions; and fittingthe navigation angle and the angle actually rotated by the agricultural robot, and verifying the correctness of a fitting function by adopting an S-shaped path. The agricultural robot platform is built, agricultural crop image information collected by the binocular camera is processed, parameters of the navigation process are extracted, and automatic navigation control over the agricultural robotis achieved.

The invention provides a drowning early warning method based on optical vision analysis. The method comprises the following steps: firstly, performing the preprocessing of denoising, enhancement and removal of uneven illumination on an image by means of median filtering, histogram equalization and homomorphic filtering; secondly, detecting and tracking swimmers simultaneously with an immune genetic algorithm to obtain position and motion track information of the swimmers; and finally, extracting behavioral characteristics of time sequence information of the swimmers through an LSTM (Long ShortTerm Memory) network, and analyzing the characteristics through a Softmax layer to obtain the drowning probability of the swimmers and give a drowning judgment. Through adoption of the method provided by the invention, intelligent visual perception of the swimmers in a swimming pool can be realized; an early warning is given about the drowning state through understanding and analysis of motion behaviors; the position information of a drowning person is given quickly and accurately; and the rescue work of lifeguards is assisted.

The invention discloses a low-quality image enhancement method under severe weather conditions, and belongs to the field of image processing. The method solves the problem that images obtained by homomorphic filtering have serious color degradation or the problem that a halo phenomenon occurs and a local part is too bright due to the Retinex algorithm when light is not uniform. According to the method, a to-be-processed image is read, the color of the image is repaired through a Retinex method into an RGB space, and a colorful improved image F is obtained; the to-be-processed image is read, the image brightness range is compressed by using a homomorphic filtering algorithm in an HSV space, and the contrast ratio is enhanced to obtain the improved brightness image Jv; the processing resultsin step 1 and step 2 are unified in the HSV space, the hue and saturation information of the image after Retinex processing is extracted, the brightness information of the image obtained after homomorphic filtering processing is extracted, and the extracted hue, the saturation information and the brightness information are fused to obtain a final image. The method is used for enhancing low-quality images.

The present invention discloses a motion vehicle detection method resisting a light effect. The method comprises the following steps: acquiring a video sequence image; extracting a motion region image in the video sequence image by using a frame difference method; analyzing a histogram of the motion region image, determining whether the image is a light unevenness image, and if yes, removing the light effect and then performing background difference processing; eliminating light unevenness of the motion region image by using a homomorphic filtering algorithm; and by using a background difference algorithm, performing motion vehicle detection on the motion region image after removing the light evenness, so as to obtain a detection result. According to the motion vehicle detection method resisting the light effect, the motion region image is extracted by using the frame difference method, and then only the motion region image is processed, so that the subsequent operation amount can be greatly reduced, and operation efficiency can be improved; and filtering is performed by using the homomorphic filtering function, thereby effectively eliminating the light unevenness effect and making the detection result more accurate.