41results about How to "Improve preprocessing speed" patented technology

The invention discloses an image enhancement and binaryzation method for a photoprint file. The image enhancement and binaryzation method for the photoprint file comprises the following steps of step one, graying input color images and obtaining graying images; step two, obtaining edge gradient of a regional image block which is corresponding to every pixel point through Sobel horizontal and vertical operator; step three, performing weight fusion improvement and LC regional enhancement and obtaining a weight fusion regional contrast value for every pixel point; step four, enabling the weight fusion regional contrast value for every pixel point to serve as a pixel value of the corresponding pixel point to obtain enhancement images of the input images. According to the image enhancement and binaryzation method for the photoprint file, the existing LC algorithm is performed on weight fusion improvement by improving the existing regional enhancement algorithm, the provided image enhancement algorithm is prior in effect compared with the existing file image enhancement and binaryzation method, has a good time performance and further improves the image preprocessing speed, and compared with the prior art, the image enhancement and binaryzation method is simple in parameter adjustment and the achievement process is simple and easy to operate.

The invention provides a determination pre-treatment device and method for a fat-soluble vitamin sample in infant milk powder. The device comprises a sealed box body, wherein a bottom plate of the sealed box body is connected with a light shielding sleeve; the light shielding sleeve is internally provided with a centrifugal pipe with a plug; the centrifugal pipe with the plug is provided with an internally-arranged sample; an opening ground plug is arranged at the opening end of the centrifugal pipe with the plug; the light shielding sleeve is in sealed connection with the bottom plate of the box body; the bottom plate of the box body, at the lower part of the light shielding sleeve, is provided with a water inlet hole for communicating an inner cavity of the light shielding sleeve and the outer part of the sealed box body; an N2 inlet is formed in one end of the sealed box body. The treatment method finished by the device comprises the step of forming the N2 inlet in the sealed box body so that the requirements of light shielding and oxygen-free environments can be met; the content and quality of vitamins in the infant formula milk powder are guaranteed; the treatment method can be used for the pre-treatment of the fat-soluble vitamins in the infant milk powder, and oxygen-free environment reaction or the pre-treatment method can be suitable for the device; and the application range is enlarged, a plurality of samples can be treated at the same time and the pre-treatment speed is increased.

The invention provides a traffic sign recognition method for a driverless car, and belongs to the technical field of image processing. According to the traffic sign recognition method, on the basis of algorithms such as a convex hull algorithm and a Hu invariant moment and transverse and longitudinal histogram scaling fast matching algorithm. Compared with existing traffic sign recognition methods, the traffic sign recognition method for the driverless car has the advantages of being large in recognition range, capable of recognizing a ban sign and an indicative sign, good in real-time performance, high in recognition accuracy and low in mistaken recognition rate.

The invention discloses a ship track abnormality detection method based on a channel model, comprising the following steps: 1. Acquiring historical AIS data in a selected time interval; Step 2. Preprocessing the historical AIS data; 3. The waypoints and routes of the waterway model are respectively extracted from the AIS data, and the waypoints and routes are respectively regarded as the vertices and edges of the graph in graph theory to form the waterway model: Step 4. Weight the waterway model according to the time element, and get The final channel model; step 5, comparing the final channel model with the track to be detected, and judging whether the track is abnormal according to whether the track to be detected appears in the final channel model. By combining the advantages of the point-based algorithm and the trajectory-based algorithm, the invention fully excavates the historical rules of navigation in the hotspot area, and obtains the behavior patterns and spatio-temporal information left by the navigation trajectory; it can perform short-term abnormal monitoring and long-term monitoring Abnormal monitoring.

The invention provides a NURBS curve interpolation fast pre-reading processing method based on an advance-retreat method. The method is characterized in that fast dispersion is carried out on a curvature smooth area of a curve through the advance-retreat method to enable parameter step increment to change quickly, and the number of curve points needing to be processed in the pretreatment process is reduced, and meanwhile, a velocity sensitive region is detected; and a curvature radius calculating function serves as an objective function, and a curvature radius minimum point of the velocity sensitive region is found through a step-length acceleration method, thereby enhancing identification and analysis processing speed of the velocity sensitive region in the forward-looking processing process and realizing fast forward. The method comprises the following specific steps: carrying out fast recognition of the velocity sensitive region through the advance-retreat method; obtaining a curvature extremum value approximation point of the velocity sensitive region; and correcting the curvature extremum value point of the velocity sensitive region, processing being finished. The method greatly reduces the number of points needing to be processed in the pretreatment process, and greatly improves real-time performance of the forward pretreatment process.

The invention discloses a drawing and coding method for large-scale complex river network massive fine regular grids of a river basin. The method comprises the steps of S1, identifying structure layers and boundary range layers of a river basin water system body; S2, obtaining river reaches of rivers of a river basin water system segment by segment; S3, drawing two-dimensional regular grids of the rivers and coding the grids in the rivers; S4, carrying out two-dimensional compilation on the two-dimensional regular grids, thereby obtaining an integrated grid; S5, carrying out one-dimensional compilation on the integrated regular grid; and S6, establishing one-dimensional grid relationships. According to the method, on the basis of an existing massive regular grid method and tool, the large-scale massive grids are increased, reduced and coarsened simply and effectively, the grids are drawn once and can be used for life, and the data support is provided for calculation of values based on the regular grids.

The invention discloses a method and a device for optical element laser preprocessing based on a total reflection principle. The method comprises that a laser beam enters the inside of a transparent optical element from a side of the transparent optical element and irradiates a processing point 1 of the front surface of the transparent optical element, the incident angle of the laser beam meets the optical total reflection condition at the processing angle 1, the laser beam is reflected to the processing point 2 of the rear surface of the element via the total reflection, the laser beam is reflected to the processing point 1 and the processing point 2 back and forth, and the laser beam is subjected to total reflection between the front surface of the inside of the element and the rear surface of the inside of the element till the laser beam exits from the other side of the transparent optical element. The laser beam is subjected to total reflection in the transparent optical element sample, the laser energy is recovered and reused, the front subsurface, the rear subsurface, the front surface, the rear surface and characteristics of the optical element are processed, and the speed of the laser preprocessing is improved.

The invention relates to a method for simplifying large-scale virtual dynamic group in the technical field of virtual reality technique. With the purpose of improving rendering speed and rendering effect of large-scale group animation, in the method of the invention, fixed grid surface spot sampling is employed and sampling point redundancy is removed to reduce information memory space, improve pre-processing speed; information retained at a sampling point list and a triangle list is used for replacing vertex information and surface information in an X file of an original model to generate X-file-like files instead of using key frame technology, so that processing complexity of group animation is simplified, different X-file-like files are created according to different LOD levels for group rendering, as a result, rendering speed is accelerated and rendering effect is improved. On the basis, the method of the invention can be employed to effective simplify the large-scale virtual dynamic group.

The invention provides a glycosylated protein sample pretreatment method based on a homogeneous reaction system. In the method, the water-soluble gold nanocluster material is dissolved in water, and acetonitrile is used for separation and enrichment of glycosylated polypeptides. The method is based on the enrichment of glycosylated polypeptides by a homogeneous system of water-soluble gold nanocluster materials. The method has high sensitivity and is suitable for pretreatment of post-translationally modified protein samples in biological samples.