43results about How to "No human interaction required" patented technology

The invention relates to medical image processing and aims to provide a multiple organ segmentation method based on a deep convolutional neural network and a regional competition model. The multiple organ segmentation method based on the deep convolutional neural network and the regional competition model comprises processes of training a three-dimensional convolutional neural network; using the trained three-dimensional convolutional neural network to learn prior probability images of liver, spleen, kidney and background in CTA volume data; determining an initial segmentation region of each tissue according to the prior probability image of each tissue; determining the probability of each pixel point belonging to each of the four tissues in the image; establishing a multiple region segmentation model based on regional competition; solving the model using the convex optimization method; and performing post-processing to obtain the contour of each organ. The invention uses the convolutional neural network to automatically and rapidly detect positions of liver, spleen and kidney at the belly, thereby obtaining the prior probability image of each organ is obtained. Then the invention uses the regional competition model, so that the contours of liver, spleen and kidney can be accurately segmented.

For multi-spectral space remote sensing data in a visible light and near-infrared range, the invention provides a relative radiometric correction technology for automatically extracting pseudo-invariant feature points for a remote sensing image. The technology mainly comprises the following steps of: for a target image with a digital number (DN) value, searching a matched reference image with surface reflectivity; pre-processing the two images; by using a canonical correlation analysis technology, searching a canonical correlation point set; screening the pseudo-invariant feature points in the canonical correlation point set; and by using the pseudo-invariant feature points, fitting a linear relation, and by using the linear relation, performing linear relative radiometric correction processing on the target image. By adoption of an algorithm, the DN value of the target image is directly converted into the surface reflectivity. The algorithm has the characteristics of simple processing flow, high processing speed and stability, and manual interaction is not required; moreover, the algorithm can be used for relative radiometric correction of the same sensors or different sensors and is particularly applicable to radiometric processing of remote sensing data, wherein the auxiliary information of the remote sensing data is lost or calibration accuracy is low, and absolute radiometric correction is not suitable for the remote sensing data.

The invention discloses a sitting posture real-time detection method based on double cameras. The method comprises the following steps of calibrating internal and external parameters of cameras; acquiring a human body sitting posture stereo image through a binocular camera; performing stereo matching for the acquired stereo image, keeping a point cloud in a set range as a potential human body part point cloud; projecting the segmented human body part point cloud in the acquired image to perform image-based human face and human eye detection; judging whether the human faces are detected; judging whether an image rotation angle is greater than a set threshold value; judging whether the human eyes are detected; calling rotation transformation R2 of a calibrated camera coordinate system relative to a vertical plane, and calculating out vertical distance between the human eyes and the cameras through a three-dimensional point cloud nearby the human eyes; adding the set distance between the binocular camera and a desk surface and the distance between the human eyes and the cameras together to obtain the vertical distance between the human eyes and the desk surface; and judging whether the vertical distance between the human eyes and the desk surface is greater than a set vertical distance threshold value. The sitting posture real-time detection method based on double cameras automatically detects human body sitting posture in real time and judges whether the sitting posture is correct or not.

The invention provides a method for acquiring a vocal organ profile in a medical image. The method comprises the following steps of: for the medical image, performing binaryzation on the lip and a background area in the medical image by using an automatic segmentation threshold of the lip and the background so as to acquire a lip profile; extracting an upper tooth profile, a lower tooth profile, an upper jaw profile and a lower jaw profile in a face range included in the lip profile; for an image area between the upper jaw profile and the lower jaw profile, acquiring a reliable edge point of a tongue profile; and fitting a tongue edge profile by the reliable edge point of the tongue profile. By the method, the vocal head part and the organ area can be automatically segmented from the image background, and the whole process is automatically completed without manual interaction.

The invention discloses a thick-cloud automatic removing method of multi-temporal remote sensing images. The method comprises the following steps: step1, collecting remote sensing satellite thick cloud images and T multi-temporal images which are in a same area with the remote sensing satellite thick cloud images; step2, detecting a thick cloud area and acquiring a thick cloud area indication template (img file=' dest_path_image 001. TIF' wi=' 40' he=' 20'/)(img file=' 153288dest_path_image 002. TIF' wi=' 40' he=' 20'/); step3, automatically selecting a reference image: automatically selecting one multi-temporal image as a reference image (img file=' dest_path_image 003. TIF' wi=' 121' he=' 44'/); step4, using a Poisson equation restoration method to remove the thick cloud and acquiring a preliminary cloud removing result (img file=' 277102dest_path_image 004.TIF' wi=' 62' he=' 25'/); step5, bringing the preliminary cloud removing result (img file=' 295873dest_path_image 004.TIF' wi=' 62' he=' 25'/) and a reference image (img file=' 708400dest_path_image 003.TIF' wi=' 121' he=' 44'/) into a variation model, removing the thick cloud again and acquiring a final cloud removing result (img file=' 313563dest_path_image 006.TIF' wi=' 137' he=' 49'/). In the invention, a reference image is determined through a root-mean-square error ((img file=' dest_path_image 007.TIF' wi=' 49' he=' 20'/)) of a thick cloud image and a plurality of multi-temporal images between gradient values. Man-machine interaction is not needed. The thick cloud and shadows are automatically removed. A pixel brightness of an original image and gradient information of the reference image are combined and a good fidelity to a pixel value is achieved.

The invention discloses a remote sensing image relative radiation correction method and system. The method comprises steps that firstly, a target image is acquired, and a reference image corresponding to the target image is determined according to the target image; secondly, the target image and the reference image are respectively pre-processed to acquire a reference image pixel point set and a target image pixel point set; thirdly, based on a nuclear typical correlation analysis method, a non-linear radiation control point is determined according to the reference image pixel point set and the target image pixel point set; and lastly, pixel points of the target image at each stage are corrected according to the non-linear radiation control point, and a surface reflectance value is acquired. The method is advantaged in that the non-linear radiation control point is determined based on the nuclear typical correlation analysis method, a problem of difficulty in guaranteeing image radiation consistency because of relatively large image radiation difference caused by relatively large time phase difference and sensor imaging geometric difference is solved, and the utilization rate of remote signaling data is substantially improved.

The invention provides a loose change exchange method, device and server. The method comprises: obtaining loose change exchange device information of the loose change exchange device by a loose changeexchange terminal corresponding to a loose change exchange account, and sending loose change exchange request information to a loose change exchange payment server; sending a start message to the loose change exchange device by the loose change exchange payment server, and starting the loose change exchange device after receiving the start message, and counting the exchange amount of the loose change to be exchanged; sending the exchange amount to the loose change exchange payment server by the loose change exchange device; and transferring the corresponding amount from a loose change collection account to the loose change exchange account by the loose change exchange payment server according to the received exchange amount, loose change exchange account information, and loose change exchange device information. According to various embodiments of the specification, electronic payment and intelligent operation of loose change exchange can be realized without the need of manual interaction. The method is convenient and fast with high efficiency and high reliability, thereby improving the user experience.

The invention discloses a fast threshold segmentation method suitable for shoe print images with poor exposure quality. Firstly, the image is pre-segmented, and the structural elements used in the morphological operation are adaptively determined according to the distribution characteristics of its 4-connected regions. Radius; secondly, use morphological opening operation to eliminate the influence of poor exposure or equipment artifacts, and then perform grayscale transformation to enhance image contrast; finally, use morphological dilation to remove residual non-shoe print pixels to obtain the final shoe print image. This method does not require human interaction, and can segment shoe print images in large batches and fully automatically. It is significantly better than manual segmentation methods and iterative segmentation methods in terms of time efficiency, and the segmentation accuracy is higher than traditional global threshold segmentation methods and iterative segmentation methods.

The invention discloses a method for efficiently solving a steady-state microdynamics equation set based on automatic combination of time integration and a Newton method. The method comprises the following steps of carrying out time integration on the initial value by adopting a time integration method based on a low-precision numerical value, automatically integrating to obtain a coarse-precisionsteady-state coverage degree by combining coverage degree sensitivity, and obtaining the steady-state coverage degree by combining with a high-precision numerical value Newton method; if single solving fails, randomly initializing the coverage in the exponential space to carry out a new round of solving, which is used for automatically solving the microcosmic kinetic equation set of the catalystsurface chemical reaction network. The method provided by the invention has higher success rate, speed and precision for solving a steady-state microdynamics equation set; meanwhile, the method is high in automation degree, automatic iterative solving can be achieved only by setting initial parameters, man-machine interaction is not needed in the process, and the algorithm process is simple and easy to master.