44results about How to "Easy to use in real time" patented technology

The invention discloses a nonlinear normalization based IQA (image quality assessment) method of a Laplace-Gaussian signal. The method comprises steps as follows: redundancy elimination expression is performed on images firstly and is completed through two processes including LOG filtering and nonlinear normalization; the two processes are used for eliminating first-order and second-order statistical redundancy in the images and reducing high-order statistical redundancy respectively; two computing methods are proposed to predict the subjective quality or the distortion degree of the images, and the two computing methods are marked as NLOG-MSE (mean square error) and NLOG-COR (correlation) respectively; according to NLOG-MSE, mean square errors between the original images subjected to redundancy elimination expression and test images is computed to obtain the distortion measurement of the test images, and according to NLOG-COR, correlation between the original image and test image in each point and redundancy elimination expression is computed to predict the image quality. The experimental result proves that the two computing methods have good prediction performance in the aspect of IQA, the NLOG-MSE method has simple computation, and the application of the NLOG-MSE method in other fields is greatly facilitated.

The invention discloses a preparation device for an intermediate compound of pregabalin, comprising a main body and a straight-through tube installed on the main body, wherein a rotating bottle is installed on the main body near the corresponding end of the straight-through tube, and the main body is close to the other end of the straight-through tube The position of the condensing pipe part that can automatically loosen the movable interface and effectively facilitate the real-time dismantling of the rotating bottle is provided. The preparation device of the pregabalin intermediate compound is different from the prior art. During the rotary evaporation process, the interfaces on the movable condenser tube parts are automatically loosened, which can effectively avoid the problem of seizing caused by the long-term tightening of the connecting parts on the condenser tube, so as to effectively maintain the condenser tube parts, and at the same time, it can effectively facilitate the real-time removal of the rotary bottle. , which facilitates the real-time use of rotary evaporator equipment.

The invention discloses a depth modeling mode decision method for 3D-HEVC (High Efficiency Video Coding) depth map coding. Classification skipping decision is performed on a depth map macro block, whether to early terminate a contour mode is judged according to the size of a prediction unit, and rate-distortion optimization is performed on a depth modeling mode to obtain an optimal depth modeling mode; the steps are as follows: starting a depth modeling mode decision process of the depth map macro block; classifying a current depth block into a near region, a middle region and a far region; skipping unnecessary depth blocks; performing early termination of the contour mode according to the size of the current prediction unit; performing a rate-distortion optimization process of a depth modeling mode decision; and determining the optimal depth modeling mode of a depth map. The depth modeling mode decision method for the 3D-HEVC depth map coding provided by the invention adopts classification skipping of blocks and early termination of the contour mode, the coding performance is better, the code rate has only a small rise, the reduction of peak signal to noise ratio can be neglected, and compared with an original 3D-HEVC coding method, the depth modeling mode decision method saves 31.33% of coding time on average, thus the method can be applied to real-time coding.

The invention discloses a video compression domain target tracking method. The method comprises the following steps: extracting compression domain information corresponding to a first frame in a videosequence from a video bit stream that is obtained through compression coding of the video sequence, decoding the compression domain information to obtain a decoded image, and carrying out manual target marking on the image to obtain a target marking diagram of the first frame; carrying out parsing on the video bit stream so that a motion vector (MV) and a block coding mode corresponding to a frame next to a current frame in the video sequence are extracted from the video bit stream, and sequentially carrying out normalization processing and preprocessing on the obtained MV to obtain a processed MV of the next frame; and obtaining a target marking diagram of the next frame through a spatial-temporal Markov model. The method provided by the invention has the advantages that the technical problems in the conventional compression domain target tracking method that block level coding processing leads to low tracking accuracy and that as the number of the frames unceasingly increases, the tracking accuracy decreases frame by frame in a continuous and free-fall manner are solved.

The invention discloses a compressed domain video action recognition method using a new spatiotemporal feature stream, comprising: extracting I frame data in a compressed video sequence, motion vector data and residual data in a P frame, and combining the motion vector data and Residual data are preprocessed. The dual-channel data of the preprocessed motion vector and the single-channel data of the preprocessed residual are respectively used as R / G / B channel fusion to construct a new spatio-temporal feature image and input into the CNN convolutional neural network model for training and testing to obtain motion recognition category scores. The extracted I-frame data and the preprocessed motion vector data are respectively input into the CNN convolutional neural network model for training and testing to obtain the recognition category score of the action, and finally fuse the three data in a ratio of 2:1:1. The recognition category score of the action gets the final action recognition result. The invention can solve the technical problems of low recognition accuracy and complicated calculation process existing in the existing video action recognition method.

The invention relates to a method for expanding the temperature measurement range of a refrigeration type temperature measurement system, which belongs to the technical field of image processing. The present invention first sets the corresponding linear interval of the detector according to the relevant parameters of the cooling type detector; then performs cumulative histogram statistics on the image to determine the minimum gray value Gl and the maximum gray value Gh in the image; judge the maximum gray value Gh and whether the minimum gray value G1 is within the range of the corresponding linear interval of the detector, if not, adjust the integration time of the detector so that the gray value in the image remains within the corresponding linear interval of the detector. The invention keeps the gray value in the image within the corresponding linear interval of the detector by effectively changing the integration time in real time, ensuring high temperature measurement accuracy and expanding the temperature measurement range of the refrigeration temperature measurement system, and at the same time the algorithm realizes It is simple, and the amount of calculation is small, which is beneficial to the real-time application of engineering.

The invention relates to a pedestrian detection method based on an on-board infrared camera. A traditional method is time consuming. According to the pedestrian detection method based on the on-board infrared camera, at first, the infrared camera is horizontally erected at the position of the car head and calibrated. Secondly, the pre-installed camera detects distance range to determine the upper and the lower boundary. And then a bounding box is selected, histograms of oriented gradients (HOG) feature of the remained candidate boxes is extracted, and a pre-trained SVM classifier is used for judging whether a bounding box is a pedestrian or not. Finally, the bounding box is judged as the pedestrian according to scale and distance clustering step 5 and the camera slides smoothly to obtain the final detection and tracking of the pedestrian. The pedestrian detection method based on the on-board infrared camera is low in time consumption and error detection, and convenient to embedded in real-time application.

The invention relates to a garbage incinerator temperature fuzzy control method using case reasoning to extract rules, and relates to the technical field of urban solid garbage incinerator temperature control. The garbage incinerator temperature fuzzy control method using case reasoning to extract rules includes the following steps: 1) using a similarity assessment method to cluster the similar data for the incinerator temperature deviation and rate of change, and the historical data output from a controller; 2) according to the maximum membership rule, generating an initial rule base; 3) using the similarity assessment method to delete the redundancy rule so as to implement reduction of the rule base; 4) calculating the similarity between the target rule and each rule in the reduction rule base to retrieve the rules; 5) according to the nearest neighbor method, reusing the rules so as to form a fuzzy control rule base; and 6) realizing stable and accurate control of the incinerator temperature through a temperature controller.

The invention relates to the technical field of urban solid waste incinerator optimal control, in particular to an automatic combustion optimized decision method for the waste incineration process. Bymeans of the optimized decision of key manipulated variables (feeder speed and fire grate speed), an automatic combustion system operates within an ideal range, and the method mainly comprises the following steps that (1) a decision case library is set up according to historical data of the incineration process; (2) a training pattern pool is constructed; (3) a stochastic allocation network is trained according to a study type pseudometric measuring criterion, and therefore a stochastic allocation network retrieval model is obtained; (4) a target case is input into the stochastic allocation network retrieval model, and K similar case solutions are obtained; (5) the average value of the K similar case solutions is obtained through case reusing, and therefore the decision value of a targetcase solution (feeder speed and fire grate speed) is obtained and output to an incineration process control system; and (6) the fourth step and the fifth step are repeated to achieve the automatic combustion real-time optimized decision process of the incineration process.