CN-121982073-A - Artificial intelligence system for big data processing

CN121982073ACN 121982073 ACN121982073 ACN 121982073ACN-121982073-A

Abstract

The invention belongs to the field of data processing, and discloses an artificial intelligence system for big data processing, which comprises an initial detection module, a motion prediction module, a detection frame generation module and a target detection module, wherein the initial detection module is used for detecting a moving target based on all pixel points in the first two frames in a frame sequence to obtain state information of the moving target in the 2 nd frame, the motion prediction module is used for predicting the state information of the moving target in the N frame based on a Kalman filtering algorithm and the state information of the moving target in the N-1 th frame, N is E [3, N ], N represents the total number of frames in the frame sequence, the detection frame generation module is used for generating a detection frame of the N frame based on the state information, and the target detection module is used for detecting the moving target in the N frame based on the detection frame to obtain the region where the moving target is located. The invention can effectively reduce the calculated amount of moving object detection and ensure the accuracy of the detection result.

Inventors

REN ZHIMING
LIANG JIANFENG
LIN ZHIHAO

Assignees

广东省科学院汕尾产业技术研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20251010

Claims (11)

1. An artificial intelligence system for big data processing is characterized by comprising an initial detection module, a motion prediction module, a detection frame generation module and a target detection module; The initial detection module is used for detecting a moving target based on all pixel points in the first two frames in the frame sequence, and obtaining state information of the moving target in the 2 nd frame; The motion prediction module is used for predicting the state information of the moving object in the nth frame based on a Kalman filtering algorithm and the state information of the moving object in the nth frame, wherein N is E [3, N ], and N represents the total number of frames in the frame sequence; the detection frame generation module is used for generating a detection frame of an nth frame based on the state information; the target detection module is used for detecting the moving target of the nth frame based on the detection frame, and obtaining the area where the moving target is located.
2. An artificial intelligence system for big data processing according to claim 1, characterized in that the status information comprises the abscissa xi, ordinate yi, width w, height h, variation of the abscissa of the center of the circumscribed frame of the moving object Variation of ordinate Variation of width w And the variation of the height h 。
3. An artificial intelligence system for big data processing according to claim 2, wherein the detecting of the moving object based on all the pixels in the first two frames in the frame sequence, the obtaining of the status information of the moving object in the 2 nd frame, comprises: respectively acquiring foreground areas corresponding to moving targets in the 1 st frame and the 2 nd frame; State information of the moving object in the 2 nd frame is calculated based on the foreground region.
4. An artificial intelligence system for big data processing according to claim 3, characterized in that for frame 1 and frame 2 the acquisition procedure of the foreground region is as follows: converting the b-th frame into a gray-scale image ,b∈{1,2}; Acquisition using image segmentation algorithm Is (are) divided into threshold values ; Acquisition of The mid-gray value is greater than the segmentation threshold A first local area composed of pixel points; screening all the obtained first local areas to obtain foreground areas representing moving targets 。
5. The artificial intelligence system for big data processing according to claim 4, wherein calculating the status information of the moving object in the 2 nd frame based on the foreground region comprises: Respectively acquiring foreground regions in a b-th frame Maximum value of abscissa of pixel point of (2) Maximum value of ordinate Minimum value of abscissa And minimum value of ordinate ; The abscissa of the center of the circumscribed frame of the moving object in the 2 nd frame is The ordinate of the center is The width of the external frame is The height of the external frame is The change amount of the abscissa is The change amount of the ordinate is The width is changed as follows The height is changed as follows 。
6. An artificial intelligence system for big data processing according to claim 2, wherein generating a detection box based on the status information comprises: acquiring the abscissa of an external frame in the state information corresponding to the nth frame Ordinate of Width of (width of) And height of ; Calculating adaptive coefficient of variation ; Based on 、 And Calculating the height and width of the detection frame; Will be And Respectively as the abscissa and the ordinate of the center of the detection frame.
7. The artificial intelligence system for big data processing according to claim 1, wherein the moving object detection is performed on the nth frame based on the detection frame to obtain the area where the moving object is located, comprising: respectively calculating a judgment value of each pixel point in the detection frame in the nth frame; Storing the pixel points with the judging values larger than the self-adaptive judging threshold value into a first set; and acquiring the region where the moving target is located based on the first set.
8. The artificial intelligence system for big data processing according to claim 7, wherein the formula of the judgment value of the pixel point is:
9. And The judgment value and the gray value of the pixel point c in the n-th frame are respectively, The gray value of the pixel point in the n-1 frame, which is the same as the coordinate of the pixel point c.
10. The artificial intelligence system for big data processing of claim 7, wherein the obtaining the region in which the moving object is located based on the first set comprises: respectively acquiring an expansion set corresponding to each pixel point in the first set; Acquiring a union of the first set and all the expansion sets; and taking the region formed by the concentrated pixel points as the region where the moving object is located.
11. The artificial intelligence system for big data processing of claim 7, wherein the obtaining the extended set corresponding to each pixel point in the first set includes: Acquiring a starting pixel point in a first set; and respectively taking each initial pixel point as a seed point to perform region growth to obtain a corresponding expansion set.

Description

Artificial intelligence system for big data processing Technical Field The invention relates to the field of data processing, in particular to an artificial intelligence system for big data processing. Background In the course of big data processing of video data, extraction of moving objects in a sequence of frames is often involved. In the conventional scheme, whether the gray value of the pixel point is changed is generally determined by calculating the difference value of the gray values of the pixel points of two adjacent frames, and after the region formed by the pixel points with the changed gray values is obtained, whether the region is a moving target can be further determined according to the area characteristics, the shape characteristics and the like of the region. But it is clearly less efficient since each comparison requires a comparison of all pixels. In order to solve the problem, in the prior art, a moving object of the first and the last frames in a frame sequence is acquired firstly, then coordinates of pixels in an area where the moving object is located are acquired in a union mode, the pixels corresponding to the acquired and concentrated coordinates form a detection area, and then gray values of the pixels of two adjacent frames are directly compared in the detection area in other frames in the frame sequence, so that the effects of reducing the number of the pixels to be compared and improving the data processing efficiency are achieved. However, this scheme has a disadvantage in that when the length of the frame sequence is long, it is possible that the moving object is partially or entirely out of the range of the detection area in some frames at the middle position of the frame sequence, and thus, an accurate moving object cannot be detected. Disclosure of Invention The invention aims to disclose an artificial intelligence system for big data processing, which solves the technical problems in the background technology. In order to achieve the above purpose, the present invention provides the following technical solutions: The invention provides an artificial intelligence system for big data processing, which comprises an initial detection module, a motion prediction module, a detection frame generation module and a target detection module; The initial detection module is used for detecting a moving target based on all pixel points in the first two frames in the frame sequence, and obtaining state information of the moving target in the 2 nd frame; The motion prediction module is used for predicting the state information of the moving object in the nth frame based on a Kalman filtering algorithm and the state information of the moving object in the nth frame, wherein N is E [3, N ], and N represents the total number of frames in the frame sequence; the detection frame generation module is used for generating a detection frame of an nth frame based on the state information; the target detection module is used for detecting the moving target of the nth frame based on the detection frame, and obtaining the area where the moving target is located. Preferably, the state information includes an abscissa xi, an ordinate yi, a width w, a height h, and a variation of an abscissa of a center of an circumscribed frame of the moving objectVariation of ordinateVariation of width wAnd the variation of the height h。 Preferably, the detecting of the moving object is performed based on all pixel points in the first two frames in the frame sequence, and the obtaining of the state information of the moving object in the 2 nd frame includes: respectively acquiring foreground areas corresponding to moving targets in the 1 st frame and the 2 nd frame; State information of the moving object in the 2 nd frame is calculated based on the foreground region. Preferably, for the 1 st and 2 nd frames, the acquisition procedure of the foreground region is as follows: converting the b-th frame into a gray-scale image ,b∈{1,2}; Acquisition using image segmentation algorithmIs (are) divided into threshold values; Acquisition ofThe mid-gray value is greater than the segmentation thresholdA first local area composed of pixel points; screening all the obtained first local areas to obtain foreground areas representing moving targets 。 Preferably, calculating the state information of the moving object in the 2 nd frame based on the foreground region includes: Respectively acquiring foreground regions in a b-th frame Maximum value of abscissa of pixel point of (2)Maximum value of ordinateMinimum value of abscissaAnd minimum value of ordinate; The abscissa of the center of the circumscribed frame of the moving object in the 2 nd frame isThe ordinate of the center isThe width of the external frame isThe height of the external frame isThe change amount of the abscissa isThe change amount of the ordinate isThe width is changed as followsThe height is changed as follows。 Preferably, generating the detection frame based on th