32 results about "Nonlinear smoothing" patented technology

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.

A horizontal direction component pixel extractor extracts a pixel of interest and a horizontal direction component pixel thereof from an input video signal. A vertical direction component pixel extractor extracts the pixel of interest and a vertical direction component pixel thereof from the input video signal. A threshold setter sets a threshold value from the pixel of interest and the vertical processing direction component pixel thereof. A non-linear smoother non-linearly smoothes the pixel of interest using a horizontal processing direction component pixel in accordance with the set threshold value, thereby generating a video signal. A flat rate calculator calculates a flat rate in a vertical direction of the pixel of interest using a vertical processing direction component pixel. A mixer mixes the input video signal and the generated video signal using the flat rate, thereby outputting a horizontal direction smoothed video signal.

The invention discloses a face detection method and system based on image on-line learning. The face detection method based on image on-line learning comprises the steps of (1) preprocessing, wherein illumination compensation and graying processing are carried out on an image to be detected, image enhancement is carried out, nonlinear smooth filtering is carried out, denoising is carried out on the image, gray levels of pixels are normalized to obtain a high-quality gray level image, and then size normalization processing and edge detection processing are carried out; (2) carrying out face gesture detection, wherein the positions of the human eyes are determined, a human face area is divided, the rotation angle of a human face in the pitching dimension, the rotation angle of the human face in the depth dimension and the rotation angle of the human face in the plane dimension are detected, and whether the human face has an expression or not is automatically judged; (3) carrying out face detection, wherein the position of the human face in the image is determined, organs of the human face are located, and gray features of the image are selected, and are transmitted to a detection template which is trained in an off-line mode to carry out judgment; (4) updating, wherein the image which is processed through detection serves as a new sample to be applied to learning of a multi-layer cascade AdaBoost classifier, and the weight of set characteristic values of the multi-layer cascade AdaBoost classifier is updated. According to the face detection method and system based on image on-line learning, the accuracy rate of face detection can be improved.

The invention discloses a linear filter and non-linear filter combined heterogeneity correction method. When a linear correction image undergoes spatial smoothing filtering to obtain an estimated image of a real scene, if the heterogeneity level is high, a mean filter with a template of 11*11 is used, or else a non-linear smoothing filter with a template of 5*5 is used, the self-adaption iteration step length of correction parameters is calculated while the filter at the edge is retained, detection and replacement of abnormal pixels (damaged elements and impulse noises) are performed, a filtering output image serves as the estimated image of the real scene, change detection is achieved by utilizing a change reference image, and iteration update is performed on a heterogeneity correction coefficient only when the difference between the correction image and the change reference image is larger than a change threshold. The linear filter and non-linear filter combined heterogeneity correction method has the advantages of large smoothing effect and quick convergence speed of a linear smoothing filter, has the edge-preservation characteristics of the non-linear smoothing filter, and is low in computing complexity and suitable for hardware circuit achievement.

The invention relates to an application method for performing satellite-borne SAR (Synthetic Aperture Radar) image ortho-rectification based on a DSM (Digital Surface Model) of median filtering, wherein the median filtering is a type of nonlinear smoothing filtering, depends on a fast sorting algorithm, and is a domain operation. The application method comprises the following steps of firstly determining a domain which takes a certain pixel as a center point, sorting the gray value of each pixel in the domain, and taking the median value as a new gray value of the center point pixel, wherein the domain is called as a window; after the window is moved up and down as well as left and right, a smoothing processing is performed on an image by utilization of a median filtering algorithm. By employment of the median filtering method, the smoothing filtering processing is performed on the DSM of an optical satellite image, and the distortion phenomenon is eliminated.

The present invention reduces a delay of an output image which is originally caused by compression or emphasis processing when compressing the dynamic range of an image or emphasizing the contrast of an image, to obtain a processed-image output on real time. The present invention provides an image processing apparatus including a non-linear smoothening unit that non-linearly smoothens an input image signal in horizontal and vertical directions while storing an edge component of the input image signal whose signal level varies, a mixing unit that mixes the image signal non-linearly smoothened by the non-linear smoothening unit with the input image signal, a correction coefficient calculation unit that calculates a gain correction coefficient, based on a mixture image signal mixed by the mixing unit, and a compression processing unit that compresses a dynamic range of the input image signal by multiplying the input image signal by the gain correction coefficient calculated by the correction coefficient calculation unit.

A heart stimulator provides for a more appropriate yet simple setting of the AV-delay. The heart stimulator comprises a stimulation pulse generator adapted to generate electric stimulation pulses and connected to a ventricular stimulation electrode for delivering electric stimulation pulses. A sensing stage connected to an electrode for picking up electric potentials inside a ventricle is adapted to sense an excitation or a contraction of a heart chamber. A memory is adapted to store parameters defining a Bezier function determining the relationship between AV-delay values and heart rate and a control unit connected to said memory, said sensing stage and to said stimulation pulse generator, is adapted to determine an actual AV-delay based on an actual intrinsic heart rate or an actual stimulation rate and a non-linear smoothing interpolation between said parameters stored in said memory.

An image processing apparatus includes a non-linear smoother for smoothing input video data with an edge component thereof preserved, and outputting smoothed video data, a subtractor for subtracting the smoothed video data from the input video data and outputting high-frequency component video data not containing the edge component, an outline extractor for extracting the edge component from the smoothed video data and outputting edge component video data, a first amplifier for varying a signal level of the edge component video data, a second amplifier for varying a signal level of the high-frequency component video data not containing the edge component, and an adder for adding video data output from the first amplifier and video data output from the second amplifier to one of the smoothed video data and the input video data.

The invention relates to a smooth control method for a power of a distributed light storage DC power supply system. A photovoltaic unit and an energy storage battery are connected to a DC bus through respective converters in parallel; the converter of the photovoltaic unit is a four-phase staggered boost converter; the maximum power p(PVmax) output is achieved by a maximum power tracking method; the converter of the energy storage battery is a four-phase staggered two-way converter; a current inner loop-power outer loop double-loop control strategy is adopted by the converter of the energy storage battery; a nonlinear smooth differentiable control method is adopted by a current inner loop; the condition that power changes of the photovoltaic unit and a load can still be quickly tracked in the presence of a load sudden change or converter parameter perturbation is ensured; and through charge-discharge control on the energy storage battery, stabilization of the power of the system is achieved. The method has relatively high anti-interference capacity on an application object with a wide power change range and large load disturbance, and has the advantages of being simple in structure, high in dynamic response speed and high in robustness.

A method for separating an audio signal into a harmonic signal component and a transient signal component is disclosed. The method includes the steps of: transferring the audio signal into a frequency space in order to obtain a transferred audio signal in dependence on frequency and time and applying a non-linear smoothing filter to the transferred audio signal over frequency to obtain a filtered transient signal T (n, k) in which the harmonic signal component is suppressed relative to the transient signal component. The method further includes applying the non-linear smoothing filter to the transferred audio signal over time to obtain a filtered harmonic signal S (n, k) in which the transient signal component is suppressed relative to the harmonic signal component and determining the harmonic signal component and the transient signal component based on the filtered harmonic signal and the filtered transient signal.

A nonlinear smooth zero return method of a carrier rocket servo mechanism comprises the steps of determining the start nonlinear return-to-zero moment and nonlinear return-to-zero moment of each servomechanism, then obtaining nonlinear return-to-zero moment of each servo mechanism and corresponding time parameters, and finally calculating to obtain a corresponding return-to-zero coefficient according to the time parameters of each servo mechanism, so that the swinging angle of an engine of each servo mechanism after correction is obtained, and the nonlinear smooth return-to-zero of the servomechanism is completed.

The invention relates to a head mounted display, and a method and device for displaying a binocular 3D video of the head mounted display, wherein the method for displaying the binocular 3D video of the head mounted display comprises the following steps of: displaying an oxyoptex or interpupillary-distance adjusting interface while obtaining an oxyoptex or interpupillary-distance adjusting signal, and performing oxyoptex or interpupillary-distance adjustment; reading video data to be displayed; generating a standard time sequence signal corresponding to the resolution of the video data, and performing non-linear smoothing treatment; judging whether the video data is 3D video data or 2D video data; dividing a 3D video data frame subjected to non-linear smoothing treatment into two parts; and respectively amplifying the two divided parts of the 3D video data frame, and respectively sending the two amplified parts of the 3D video data frame to two display screens to display. By means of the scheme disclosed by the invention, binocular oxyoptex and interpupillary-distance adjustment of a binocular optical module and binocular 3D video display of the head mounted display can be completed rapidly and conveniently; and thus, users having different oxyoptex and interpupillary distances can obtain perfect viewing effects.

An image processing apparatus includes a non-linear smoother for smoothing input video data with an edge component thereof preserved, and outputting smoothed video data, a subtractor for subtracting the smoothed video data from the input video data and outputting high-frequency component video data not containing the edge component, an outline extractor for extracting the edge component from the smoothed video data and outputting edge component video data, a first amplifier for varying a signal level of the edge component video data, a second amplifier) for varying a signal level of the high-frequency component video data not containing the edge component, and an adder for adding video data output from the first amplifier and video data output from the second amplifier to one of the smoothed video data and the input video data.

A system for enhancing an input image including receiving an input image and filtering the input image with a plurality of non-linear smoothing filters providing a respective plurality of filtered outputs. The system processes a plurality of the filtered outputs with respect to at least one of another of the filtered outputs and the input image to determine a plurality of detail layers. The system filters the plurality of detail layers with a plurality of non-linear smoothing filters providing a respective plurality of smoothed layers. The system adjusts the plurality of smoothed layers in such a manner that regions closer to an edge are enhanced to a lesser extent than regions farther from an edge and combining the adjusted the smoothed layers to provide an enhanced output image.

A system for enhancing an input image including receiving an input image and filtering the input image with a plurality of non-linear smoothing filters providing a respective plurality of filtered outputs. The system processes a plurality of the filtered outputs with respect to at least one of another of the filtered outputs and the input image to determine a plurality of detail layers. The system filters the plurality of detail layers with a plurality of non-linear smoothing filters providing a respective plurality of smoothed layers. The system adjusts the plurality of smoothed layers in such a manner that regions closer to an edge are enhanced to a lesser extent than regions farther from an edge and combining the adjusted the smoothed layers to provide an enhanced output image.

The invention provides a self-cruise adjusting device and method for automatic exposure. According to concept of Shannon's entropy of information, the information entropy value of the scene image is calculated, and exposure is adjusted nonlinearly and smoothly, so that the scene image with optimal exposure is obtained. Compared with traditional single-brightness target exposure, the method has better scene adaptability. Compared with other multi-brightness target exposure methods, the method has more natural coherence and stability. The invention further provides a hardware accelerator matchedwith the self-cruise adjusting device and method for automatic exposure.

The invention relates to a method for evaluating an image dataset obtained by a radiation-based image acquisition device. A scatter background dataset is determined as a function of the image data. The image dataset is corrected pixel by pixel by multiplying the image dataset with the inverse of a function dependent on the quotient of the scatter background data and the image data at a respective pixel. The function is a nonlinear, smooth function determined by a coefficient and having positive derivatives. The absolute value of the function is one for the value zero. The image acquisition parameter dependent coefficient is determined by an optimization process.

The invention discloses a method for controlling the movement of a large-screen television mouse by a mobile phone, which comprises the following steps of: performing nonlinear smooth change on an input original movement distance through an algorithm, so that a user can obtain smaller mouse movement when the movement distance is smaller, and can obtain larger mouse movement when the touch movement distance is larger. And secondly, the decimal part which is not moved at this time of rounding data is compensated in the next movement. The operation intention of the user is reflected more accurately, and better user experience is obtained.

The invention relates to an unmanned ship navigation method based on a millimeter-wave radar, and the method comprises the following steps: S1, radar point cloud data processing: sensing obstacles around an unmanned ship through the millimeter-wave radar, and issuing relatively stable obstacle list data; S2, local map real-time updating: subscribing the obstacle list and putting the obstacle list into an obstacle database to form a stable local map; s3, transverse path planning: subscribing positioning and attitude data, a local map and a local target waypoint, and calculating a safe and feasible path within prediction time by performing continuous two-time sampling evaluation of safety margin on a heading space; s4, longitudinal speed planning: subscribing positioning and attitude data, and using a nonlinear smooth function to give a safe navigational speed within prediction time; and S5, track tracking: issuing the calculated target waypoint and target navigational speed to a control unit for track tracking. The method has the advantages of being real-time, effective, universal, practical, stable and safe.

A system for enhancing an input image including receiving an input image and filtering the input image with a plurality of non-linear smoothing filters providing a respective plurality of filtered outputs. The system processes a plurality of the filtered outputs with respect to at least one of another of the filtered outputs and the input image to determine a plurality of detail layers. The system filters the plurality of detail layers with a plurality of non-linear smoothing filters providing a respective plurality of smoothed layers. The system adjusts the plurality of smoothed layers in such a manner that regions closer to an edge are enhanced to a lesser extent than regions farther from an edge and combining the adjusted the smoothed layers to provide an enhanced output image.