CN113256559A - Rapid detection method for high-altitude object throwing - Google Patents
Rapid detection method for high-altitude object throwing Download PDFInfo
- Publication number
- CN113256559A CN113256559A CN202110398358.4A CN202110398358A CN113256559A CN 113256559 A CN113256559 A CN 113256559A CN 202110398358 A CN202110398358 A CN 202110398358A CN 113256559 A CN113256559 A CN 113256559A
- Authority
- CN
- China
- Prior art keywords
- altitude
- frame
- image
- frame image
- parabola
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
Abstract
The invention provides a high-altitude parabolic rapid detection method, which comprises the following steps: acquiring continuous frame images from a real-time video acquired from a floor, and segmenting a region to be detected by using a depth learning model for a first frame image; detecting a moving object in each frame of image, analyzing the motion condition of the moving object in the frame of image by depending on the previous frame of image, and preliminarily judging whether the moving object has a high-altitude parabolic region; analyzing the characteristics of the moving object for the frame image with the high-altitude parabolic region; judging whether a high-altitude parabola exists or not according to the analysis result aiming at the continuous m frames of images, acquiring a trajectory graph judged as the high-altitude parabola, inputting the trajectory graph into a deep learning model for detection, and verifying whether the trajectory graph is the high-altitude parabola or not; and taking the next frame image of the m frame images as the last frame of the next continuous m frame images, and deleting the first frame of the previous m frame images. The invention greatly improves the detection speed on the basis of ensuring the detection precision.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a high-altitude parabolic rapid detection method.
Background
The high-altitude parabolic model not only is an uneventful behavior, but also has great social hazard. Individuals who practice this act, on a light basis, undertake economic compensation and, on a heavy basis, undertake criminal liability and also incur losses to innocent misleaders or victims. The method for detecting the high-altitude object throwing aims at the problems of potential safety hazards and the like caused by the high-altitude object throwing. At present, the method for detecting the high-altitude parabola needs to detect the moving object first, then track the moving object and finally judge whether the moving object is the high-altitude parabola or not. Due to the fact that the image scene acquired by the high-altitude parabolic camera is large, the size of the moving object is small, the moving object is multiple in size, the detection speed of the moving object tracking is low, the moving object tracking is easy to lose, and the misjudgment rate is increased. The high-altitude parabolic motion speed is high, the occurrence time is short, and the detection speed is inevitably required to be greatly improved on the basis of ensuring the detection precision.
Disclosure of Invention
The invention provides a high-altitude parabolic rapid detection method, aiming at solving the problem of low high-altitude parabolic detection speed in the prior art.
The technical scheme of the invention is realized as follows:
a method for rapidly detecting a high altitude parabola includes: acquiring continuous frame images from a real-time video acquired from a floor, and segmenting a region to be detected by using a deep learning model for a first frame image; detecting a moving object in each frame of image, analyzing the motion condition of the moving object in the frame of image by depending on the previous frame of image, and preliminarily judging whether the moving object has a high-altitude parabolic region; analyzing the characteristics of the moving object for the frame image with the high-altitude parabolic region; judging whether a high-altitude parabola exists or not according to the analysis result aiming at the continuous m frames of images, acquiring a trajectory graph judged as the high-altitude parabola, inputting the trajectory graph into a deep learning model for detection, and verifying whether the trajectory graph is the high-altitude parabola or not; and taking the next frame image of the m frames of images as the last frame of the next continuous m frames of images, and deleting the first frame of the previous m frames.
Preferably, the method for rapidly detecting the high altitude parabola further comprises the steps of judging the current weather condition of the first frame image by using a deep learning model when the region to be detected is segmented; the corresponding frame image can be used for judging the current weather condition by using the deep learning model at regular time, and the area to be detected is segmented.
Preferably, in the method for rapidly detecting a high altitude parabola, the specific way of preliminarily determining whether the region with the high altitude parabola exists is as follows: dividing each frame image into n vertical bar blocks, dividing each vertical bar block into a plurality of image slices, and counting and recording the number of moving pixel points in each image slice;
comparing the moving pixel point number of each image slice of each vertical block of the frame image with the moving pixel point number of the corresponding image slice of the corresponding vertical block of the previous frame image from the second frame image, comparing the positive difference value of the comparison with a given pixel point number threshold, preliminarily screening the image slices with a large number of moving pixel points and larger than the pixel point number threshold, and counting the number of the image slices meeting the conditions in each vertical block; and giving an image slice number threshold value, further screening vertical bar blocks larger than the image slice number threshold value, and preliminarily judging the areas with high altitude parabolas by using the vertical bar blocks screened out by the current frame.
Preferably, in the method for rapidly detecting a high altitude parabola, the specific way of analyzing the characteristics of the moving object for the frame image having the high altitude parabola region is as follows: grouping the moving pixel points of all image pieces screened from each vertical bar block which is preliminarily judged to have high-altitude parabolas in the frame image according to the gray value, setting the grouping number k, calculating the length h =256/k of each combined gray value according to the value range of 0-255 of the gray value, and sequentially determining the grouping, namely 0-h-1, h-2 h-1, 2 h-3 h-1 and … until the grouping is finished; calculating Y coordinate mean values of moving pixel points of all groups of each vertical bar block, and recording the Y coordinate mean values of the moving pixel points of different gray level combinations of each vertical bar block; each group of each vertical bar block in the frame image which is not preliminarily judged to have a high altitude parabola is recorded.
Preferably, in the method for rapidly detecting a high altitude parabola, the specific way of judging whether the high altitude parabola exists or not according to the analysis result for the continuous m-frame images is as follows: comparing the Y coordinate mean value of the moving pixel points with different gray combinations of each vertical block of the current frame image with the Y coordinate mean value of the moving pixel points with corresponding gray combinations of the corresponding vertical blocks of the previous frame image from the third frame image, continuously comparing the n vertical blocks with the m frame image, and counting the image frame number with the increased Y coordinate mean value of the moving pixel points with different gray combinations of each vertical block in all comparisons; setting a frame number threshold, and screening out vertical bar blocks of m-frame images with the image frame number greater than the frame number threshold and corresponding gray level combinations thereof; calculating the difference value of the Y coordinate mean value of the moving pixel points of the vertical bar blocks of the next frame image and the Y coordinate mean value of the moving pixel points of the vertical bar blocks of the previous frame image of the same gray scale combination for the gray scale combination of the vertical bar blocks of each group of m frame images, wherein the difference value is a quadratic function related to the motion time, calculating quadratic differential on the motion time of the difference value to obtain the Y-direction acceleration of the moving pixel points of the vertical bar blocks on the two adjacent frame images, and obtaining m-2 accelerations of the moving pixel points of the vertical bar blocks of the m frame images; and (3) giving an acceleration threshold, screening m-2 vertical bar blocks of the m-frame image with the acceleration larger than the acceleration threshold, and judging that high-altitude parabolas exist in the vertical bar blocks of the m-frame image.
Preferably, in the method for rapidly detecting a high altitude parabola, the specific way of acquiring the trajectory graph determined as the high altitude parabola is as follows: continuously judging high altitude parabolas of m frames of images, continuously updating a next frame of image into a new m frames of image, judging that the image continuously having the high altitude parabolas is s frames, and setting a frame number threshold value t, wherein m = < s < = t; for the vertical bar blocks of the s-frame image which is judged to have the high-altitude parabola, the coordinate positions of moving pixel points in the corresponding gray level combination of the vertical bar blocks of each frame image in the vertical bar blocks are obtained, the moving pixel points are track points of the high-altitude parabola in the track map, and the size of the track map is the same as that of any frame image; and pixel values of all pixel positions of the non-high-altitude parabolic object in the last frame image of the s frame image are equally corresponding to the locus map, the pixel value of each locus point in the corresponding frame image and the pixel value of the background image of the moving object in the corresponding frame image are weighted and averaged, the pixel value of the locus point in the locus map is obtained, and the locus map which is judged to be the high-altitude parabolic object is obtained.
Preferably, in the method for rapidly detecting a high-altitude parabola, a deep learning model is input for detection, and a specific way for verifying whether the high-altitude parabola is detected is as follows: establishing a two-classification deep learning model with and without high-altitude parabolas, and inputting all the trajectory graphs judged as the high-altitude parabolas into the two-classification deep learning model for verification.
Preferably, in the method for rapidly detecting the high altitude parabola, for s frames of images continuously having the high altitude parabola, the s frame may be the (m + 1) th frame or the (m + 2) th frame, and if s = t, a moving pixel point of the high altitude parabola obtained according to the s frames of images is a track point of the current high altitude parabola; if s < t, the corresponding position of the next frame image of the s frame does not have a high-altitude parabola, setting a high-altitude parabola disappearance threshold, if the continuous frame number of the non-high-altitude parabola is larger than the high-altitude parabola disappearance threshold, judging that the track point corresponding to the s frame image is the track point of the current high-altitude parabola, if the continuous frame number of the non-high-altitude parabola is smaller than or equal to the high-altitude parabola disappearance threshold, judging that the high-altitude parabola still exists from the s frame to the current frame, and continuously recording the high-altitude parabola track point.
Preferably, the method for rapidly detecting a high altitude parabola further includes: judging and verifying the danger of the high-altitude object, wherein the specific judging mode is as follows: and extracting m-2 accelerated speeds of the vertical bar blocks of the m frames of images corresponding to the high-altitude parabola, giving a danger threshold, and judging that the high-altitude parabola has danger if the m-2 accelerated speeds are larger than the danger threshold.
Preferably, the method for rapidly detecting a high altitude parabola further includes: and early warning is carried out on the high-altitude object which is judged to be dangerous.
The invention has the beneficial effects that: the invention relates to a high-altitude parabolic rapid detection method, which detects moving objects in each frame of image, analyzes the motion condition of the moving objects in the frame of image by depending on the previous frame of image, and preliminarily judges whether a high-altitude parabolic region exists or not; analyzing the characteristics of the moving object for the frame image with the high-altitude parabolic region; and judging whether a high-altitude parabola exists or not according to the analysis result aiming at the continuous m frames of images, and finally verifying whether the high-altitude parabola exists or not. The method for rapidly detecting the high-altitude object throwing completely avoids tracking the moving object after detecting the moving object, greatly simplifies algorithm operation, and improves detection speed in a crossing mode. And the whole detection method judges whether the object is a high-altitude object for many times according to the actual motion condition of the moving object, and finally verifies, so that the high-altitude object detection precision is improved. The invention greatly improves the detection speed on the basis of ensuring the detection precision.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic diagram of a method for rapidly detecting a high altitude parabola according to the present invention;
FIG. 2 is a diagram illustrating an initial determination of a region with high altitude parabola.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example (b): a method for rapidly detecting a high altitude parabola as shown in fig. 1 comprises: acquiring continuous frame images from a real-time video acquired from a floor, and segmenting a region to be detected by using a deep learning model for a first frame image; detecting a moving object in each frame of image, analyzing the motion condition of the moving object in the frame of image by depending on the previous frame of image, and preliminarily judging whether the moving object has a high-altitude parabolic region; analyzing the characteristics of the moving object for the frame image with the high-altitude parabolic region; judging whether a high-altitude parabola exists or not according to the analysis result aiming at the continuous m frames of images, acquiring a trajectory graph judged as the high-altitude parabola, inputting the trajectory graph into a deep learning model for detection, and verifying whether the trajectory graph is the high-altitude parabola or not; and taking the next frame image of the m frames of images as the last frame of the next continuous m frames of images, and deleting the first frame of the previous m frames.
When the first frame image is divided into the regions to be detected by the deep learning model, the current weather condition is judged by the other deep learning model, the weather conditions are cloudy days, rainy days, snowy days and the like, and the false alarm interference caused by the weather is greatly avoided. Rain falls, snowflakes fall continuously, the image processing difficulty is increased, and false alarm is increased. The area to be detected is a high building, and the purpose is to eliminate the interference of the surrounding environment, flowers, plants and trees. The corresponding frame images can be used for judging the current weather condition by using the deep learning model at regular time, the area to be detected is segmented, different interference backgrounds can be removed in time, and the detection precision and efficiency are improved.
And detecting a moving object in each frame of image, analyzing the motion condition of the moving object in the frame of image by depending on the previous frame of image, and preliminarily judging the region with the high-altitude parabola. The specific way of judging is as follows: dividing each frame image into n vertical bar blocks, dividing each vertical bar block into a plurality of image slices, and counting and recording the number of moving pixel points in each image slice; comparing the moving pixel point number of each image slice of each vertical block of the frame image with the moving pixel point number of the corresponding image slice of the corresponding vertical block of the previous frame image from the second frame image, comparing the positive difference value of the comparison with a given pixel point number threshold, preliminarily screening the image slices with a large number of moving pixel points and larger than the pixel point number threshold, and counting the number of the image slices meeting the conditions in each vertical block; and giving an image slice number threshold value, further screening vertical bar blocks larger than the image slice number threshold value, and preliminarily judging the areas with high altitude parabolas by using the vertical bar blocks screened out by the current frame.
When high-altitude parabolas appear, the number of moving pixel points of the current frame image position of the high-altitude parabolas is suddenly increased compared with the corresponding position of the previous frame image, and for the position of a moving object interfered in the image or the position without the moving object, the number of moving pixel points of the corresponding positions of the previous frame image and the next frame image is basically unchanged, so that the region with the high-altitude parabolas is preliminarily judged, the interference of noise is reduced, high-altitude parabolas are detected only aiming at the screened region, and the detection speed is improved.
From the second frame image, the characteristics of the moving object are analyzed for the frame image having the high-altitude parabolic region, and the next frame is directly detected for the frame image without the high-altitude parabolic region. The specific way to analyze the characteristics of the moving object for the frame image with the high-altitude parabolic region is as follows: grouping the moving pixel points of all image pieces screened from each vertical bar block which is preliminarily judged to have high-altitude parabolas in the frame image according to the gray value, setting the grouping number k, calculating the length h =256/k of each combined gray value according to the value range of 0-255 of the gray value, and sequentially determining the grouping, namely 0-h-1, h-2 h-1, 2 h-3 h-1 and … until the grouping is finished; preliminarily judging that moving pixel points of all image slices screened from each vertical bar block with high altitude parabolic shape are totally divided into k groups, calculating Y coordinate mean values of the moving pixel points of all groups of each vertical bar block, and recording the Y coordinate mean values of the moving pixel points of different gray level combinations of each vertical bar block; each group of each vertical bar block in the frame image that is not preliminarily judged to have a high altitude parabola is recorded. Here, each group of each vertical bar block which is not preliminarily judged to have a high-altitude parabola in the frame image can be recorded in any manner according to actual conditions, and the frame image is recorded without the high-altitude parabola in the vertical bar. One way of recording is illustrated: the Y-coordinate mean of each group of each vertical bar block in the frame image that is not preliminarily determined to have a high-altitude parabola is set to a fixed negative number indicating that the frame image does not have a high-altitude parabola within the vertical bar.
The moving pixel points of all image slices screened out from each vertical bar block which is preliminarily judged to have the high-altitude parabolic object are grouped according to the gray value, the grouping formula mainly distinguishes different moving objects which may be the high-altitude parabolic objects in the frame image, the gray value of the same high-altitude parabolic object does not change too much in the motion process, the gray value of different objects may have larger difference, and therefore the moving objects which are not the high-altitude parabolic objects are removed or the high-altitude parabolic objects with different gray values are distinguished.
Starting from the second frame image, each frame image is processed and analyzed as above, continuous m frame images are processed and analyzed all the time, and after the last frame image is processed, whether high-altitude parabolic exists or not is judged according to the analysis result aiming at the continuous m frame images, wherein the judgment specific mode is as follows: comparing the Y coordinate mean value of the moving pixel points with different gray combinations of each vertical block of the current frame image with the Y coordinate mean value of the moving pixel points with corresponding gray combinations of the corresponding vertical blocks of the previous frame image from the third frame image, continuously comparing the n vertical blocks with the m frame image, and counting the image frame number with the increased Y coordinate mean value of the moving pixel points with different gray combinations of each vertical block in all comparisons; setting a frame number threshold, and screening out vertical bar blocks of m-frame images with the image frame number greater than the frame number threshold and corresponding gray level combinations thereof; calculating the difference value of the Y coordinate mean value of the moving pixel points of the vertical bar blocks of the next frame image and the Y coordinate mean value of the moving pixel points of the vertical bar blocks of the previous frame image of the same gray scale combination for the gray scale combination of the vertical bar blocks of each group of m frame images, wherein the difference value is a quadratic function related to the motion time, calculating quadratic differential on the motion time of the difference value to obtain the Y-direction acceleration of the moving pixel points of the vertical bar blocks on the two adjacent frame images, and obtaining m-2 accelerations of the moving pixel points of the vertical bar blocks of the m frame images; and (3) giving an acceleration threshold, screening m-2 vertical bar blocks of the m-frame image with the acceleration larger than the acceleration threshold, and judging that high-altitude parabolas exist in the vertical bar blocks of the m-frame image. The motion of the high-altitude object is a continuous process, and in order to accurately analyze whether the high-altitude object exists, the motion condition of the moving object of the continuous frame images needs to be analyzed.
Continuously processing m frames of images to obtain the difference value of m-1Y coordinate mean values, setting the time interval between two adjacent frames of images as unit time dt for the motion displacement between two corresponding adjacent frames of images, and according to the relation of displacement and acceleration time, using the quadratic differential of the displacement to represent the motion acceleration of the object, wherein the total number of the acceleration is m-2.
The specific way of acquiring the trajectory graph determined as the high altitude parabola is as follows: continuously judging high altitude parabolas of m frames of images, continuously updating a next frame of image into a new m frames of image, judging that the image continuously having the high altitude parabolas is s frames, and setting a frame number threshold value t, wherein m = < s < = t; for the vertical bar blocks of the s-frame image which is judged to have the high-altitude parabola, the coordinate positions of moving pixel points in the corresponding gray level combination of the vertical bar blocks of each frame image in the vertical bar blocks are obtained, the moving pixel points are track points of the high-altitude parabola in the track map, and the size of the track map is the same as that of any frame image; and pixel values of all pixel positions of the non-high-altitude parabolic object in the last frame image of the s frame image are equally corresponding to the locus map, the pixel value of each locus point in the corresponding frame image and the pixel value of the background image of the moving object in the corresponding frame image are weighted and averaged, the pixel value of the locus point in the locus map is obtained, and the locus map which is judged to be the high-altitude parabolic object is obtained. The weight coefficient for weighted averaging can be set according to specific situations. The weighted average is carried out by fusing the pixel values of the pixel points which are determined as high-altitude parabolic track points in each frame of the s-frame image, the fusion aims to highlight the characteristics of the track points in the track image, classification and verification are facilitated by using a deep learning model, and for misjudged interference moving pixel points such as fluttering leaves, the pixel values are basically consistent with those of the current frame image after weighted average due to small position change, the characteristics of the interference moving objects cannot be highlighted in the track image, and the misjudged moving pixel points can be removed when the verification is carried out by using the deep learning model, so that the misjudgment rate is reduced.
For s frames of images with continuous high-altitude parabolas, the s frame may be the (m + 1) th frame or the (m + 2) th frame, if s = t, a high-altitude parabolic moving pixel point obtained according to the s frames of images is a track point of the current high-altitude parabolas; if s < t, the corresponding position of the next frame image of the s frame does not have a high-altitude parabola, setting a high-altitude parabola disappearance threshold, if the continuous frame number of the non-high-altitude parabola is larger than the high-altitude parabola disappearance threshold, judging that the track point corresponding to the s frame image is the track point of the current high-altitude parabola, if the continuous frame number of the non-high-altitude parabola is smaller than or equal to the high-altitude parabola disappearance threshold, judging that the high-altitude parabola still exists from the s frame to the current frame, and continuously recording the high-altitude parabola track point.
Inputting the trajectory graph of the high-altitude parabola into a deep learning model for detection, and verifying whether the trajectory graph is the high-altitude parabola or not in a specific mode that: establishing a two-classification deep learning model with and without high-altitude parabolas, and inputting all the trajectory graphs judged as the high-altitude parabolas into the two-classification deep learning model for verification.
The invention also discloses a high-altitude parabolic rapid detection method, which comprises the following steps: and judging and verifying the dangerousness of the high-altitude object, and early warning the high-altitude object judged to be dangerous. The specific way for judging and verifying the risk of the high-altitude object is as follows: and extracting m-2 accelerated speeds of the vertical bar blocks of the m frames of images corresponding to the high-altitude parabola, giving a danger threshold, and judging that the high-altitude parabola has danger if the m-2 accelerated speeds are larger than the danger threshold.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. A high altitude parabola rapid detection method is characterized by comprising the following steps: acquiring continuous frame images from a real-time video acquired from a floor, and segmenting a region to be detected by using a deep learning model for a first frame image; detecting a moving object in each frame of image, analyzing the motion condition of the moving object in the frame of image by depending on the previous frame of image, and preliminarily judging whether the moving object has a high-altitude parabolic region; analyzing the characteristics of the moving object for the frame image with the high-altitude parabolic region; judging whether a high-altitude parabola exists or not according to the analysis result aiming at the continuous m frames of images, acquiring a trajectory graph judged as the high-altitude parabola, inputting the trajectory graph into a deep learning model for detection, and verifying whether the trajectory graph is the high-altitude parabola or not; and taking the next frame image of the m frames of images as the last frame of the next continuous m frames of images, and deleting the first frame of the previous m frames.
2. The high-altitude parabolic rapid detection method according to claim 1, further comprising judging a current weather condition when the first frame image is used for segmenting the region to be detected by using a deep learning model; the corresponding frame image can be used for judging the current weather condition by using the deep learning model at regular time, and the area to be detected is segmented.
3. The method for rapidly detecting the high altitude parabola as claimed in claim 1, wherein the specific way of preliminarily judging whether the high altitude parabola exists in the area is as follows: dividing each frame image into n vertical bar blocks, dividing each vertical bar block into a plurality of image slices, and counting and recording the number of moving pixel points in each image slice;
comparing the moving pixel point number of each image slice of each vertical block of the frame image with the moving pixel point number of the corresponding image slice of the corresponding vertical block of the previous frame image from the second frame image, comparing the positive difference value of the comparison with a given pixel point number threshold, preliminarily screening the image slices with a large number of moving pixel points and larger than the pixel point number threshold, and counting the number of the image slices meeting the conditions in each vertical block; and giving an image slice number threshold value, further screening vertical bar blocks larger than the image slice number threshold value, and preliminarily judging the areas with high altitude parabolas by using the vertical bar blocks screened out by the current frame.
4. The method for rapidly detecting a high altitude parabola as claimed in claim 3, wherein the specific way of analyzing the characteristics of the moving object for the frame image with the high altitude parabola region is as follows: grouping the moving pixel points of all image pieces screened from each vertical bar block which is preliminarily judged to have high-altitude parabolas in the frame image according to the gray value, setting the grouping number k, calculating the length h =256/k of each combined gray value according to the value range of 0-255 of the gray value, and sequentially determining the grouping, namely 0-h-1, h-2 h-1, 2 h-3 h-1 and … until the grouping is finished; calculating Y coordinate mean values of moving pixel points of all groups of each vertical bar block, and recording the Y coordinate mean values of the moving pixel points of different gray level combinations of each vertical bar block; each group of each vertical bar block in the frame image which is not preliminarily judged to have a high altitude parabola is recorded.
5. The method for rapidly detecting the high altitude parabola according to claim 4, wherein the specific way of judging whether the high altitude parabola exists or not according to the analysis result for the continuous m frames of images is as follows: comparing the Y coordinate mean value of the moving pixel points with different gray combinations of each vertical block of the current frame image with the Y coordinate mean value of the moving pixel points with corresponding gray combinations of the corresponding vertical blocks of the previous frame image from the third frame image, continuously comparing the n vertical blocks with the m frame image, and counting the image frame number with the increased Y coordinate mean value of the moving pixel points with different gray combinations of each vertical block in all comparisons; setting a frame number threshold, and screening out vertical bar blocks of m-frame images with the image frame number greater than the frame number threshold and corresponding gray level combinations thereof; calculating the difference value of the Y coordinate mean value of the moving pixel points of the vertical bar blocks of the next frame image and the Y coordinate mean value of the moving pixel points of the vertical bar blocks of the previous frame image of the same gray scale combination for the gray scale combination of the vertical bar blocks of each group of m frame images, wherein the difference value is a quadratic function related to the motion time, calculating quadratic differential on the motion time of the difference value to obtain the Y-direction acceleration of the moving pixel points of the vertical bar blocks on the two adjacent frame images, and obtaining m-2 accelerations of the moving pixel points of the vertical bar blocks of the m frame images; and (3) giving an acceleration threshold, screening m-2 vertical bar blocks of the m-frame image with the acceleration larger than the acceleration threshold, and judging that high-altitude parabolas exist in the vertical bar blocks of the m-frame image.
6. The high-altitude parabolic rapid detection method according to claim 5, wherein the specific way of obtaining the trajectory graph determined as the high-altitude parabolic is as follows: continuously judging high altitude parabolas of m frames of images, continuously updating a next frame of image into a new m frames of image, judging that the image continuously having the high altitude parabolas is s frames, and setting a frame number threshold value t, wherein m = < s < = t; for the vertical bar blocks of the s-frame image which is judged to have the high-altitude parabola, the coordinate positions of moving pixel points in the corresponding gray level combination of the vertical bar blocks of each frame image in the vertical bar blocks are obtained, the moving pixel points are track points of the high-altitude parabola in the track map, and the size of the track map is the same as that of any frame image; and pixel values of all pixel positions of the non-high-altitude parabolic object in the last frame image of the s frame image are equally corresponding to the locus map, the pixel value of each locus point in the corresponding frame image and the pixel value of the background image of the moving object in the corresponding frame image are weighted and averaged, the pixel value of the locus point in the locus map is obtained, and the locus map which is judged to be the high-altitude parabolic object is obtained.
7. The method for rapidly detecting the high altitude parabola according to claim 6, wherein a deep learning model is input for detection, and the specific way for verifying whether the high altitude parabola is detected is as follows: establishing a two-classification deep learning model with and without high-altitude parabolas, and inputting all the trajectory graphs judged as the high-altitude parabolas into the two-classification deep learning model for verification.
8. The high-altitude parabolic quick detection method according to claim 6, characterized in that for s frames of images with continuous high-altitude parabolic features, the s frame may be the (m + 1) th frame or the (m + 2) th frame, if s = t, the high-altitude parabolic moving pixel points obtained according to the s frames of images are track points of the current high-altitude parabolic feature; if s < t, the corresponding position of the next frame image of the s frame does not have a high-altitude parabola, setting a high-altitude parabola disappearance threshold, if the continuous frame number of the non-high-altitude parabola is larger than the high-altitude parabola disappearance threshold, judging that the track point corresponding to the s frame image is the track point of the current high-altitude parabola, if the continuous frame number of the non-high-altitude parabola is smaller than or equal to the high-altitude parabola disappearance threshold, judging that the high-altitude parabola still exists from the s frame to the current frame, and continuously recording the high-altitude parabola track point.
9. The method for rapidly detecting the high altitude parabola according to claim 7, further comprising: judging and verifying the danger of the high-altitude object, wherein the specific judging mode is as follows: and extracting m-2 accelerated speeds of the vertical bar blocks of the m frames of images corresponding to the high-altitude parabola, giving a danger threshold, and judging that the high-altitude parabola has danger if the m-2 accelerated speeds are larger than the danger threshold.
10. The method for rapidly detecting the high altitude parabola according to claim 7, further comprising: and early warning is carried out on the high-altitude object which is judged to be dangerous.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110398358.4A CN113256559A (en) | 2021-04-14 | 2021-04-14 | Rapid detection method for high-altitude object throwing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110398358.4A CN113256559A (en) | 2021-04-14 | 2021-04-14 | Rapid detection method for high-altitude object throwing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113256559A true CN113256559A (en) | 2021-08-13 |
Family
ID=77220721
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110398358.4A Pending CN113256559A (en) | 2021-04-14 | 2021-04-14 | Rapid detection method for high-altitude object throwing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113256559A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114596531A (en) * | 2022-03-31 | 2022-06-07 | 深圳市海清视讯科技有限公司 | High-altitude parabolic detection method, device, equipment and storage medium |
-
2021
- 2021-04-14 CN CN202110398358.4A patent/CN113256559A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114596531A (en) * | 2022-03-31 | 2022-06-07 | 深圳市海清视讯科技有限公司 | High-altitude parabolic detection method, device, equipment and storage medium |
| CN114596531B (en) * | 2022-03-31 | 2022-08-05 | 深圳市海清视讯科技有限公司 | High-altitude parabolic detection method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108446630B (en) | Intelligent monitoring method for airport runway, application server and computer storage medium | |
| CN115691026B (en) | Intelligent early warning monitoring management method for forest fire prevention | |
| KR101353952B1 (en) | Method for detecting wildfire smoke using spatiotemporal bag-of-features of smoke and random forest | |
| CN110874592B (en) | Forest fire smoke image detection method based on total bounded variation | |
| CN101403676B (en) | Insulator hydrophobicity rank amalgamation judging method based on D-S evidence theory | |
| CN107085714B (en) | Forest fire detection method based on video | |
| CN110781723B (en) | Group abnormal behavior identification method | |
| US9613432B2 (en) | Fire detection system and method employing digital images processing | |
| CN111626170B (en) | Image recognition method for railway side slope falling stone intrusion detection | |
| CN111626169B (en) | Image-based railway dangerous falling rock size judgment method | |
| CN104463253B (en) | Passageway for fire apparatus safety detection method based on adaptive background study | |
| CN111860143B (en) | Real-time flame detection method for inspection robot | |
| CN115035182B (en) | Landslide hazard early warning method and system | |
| CN112906481A (en) | Method for realizing forest fire detection based on unmanned aerial vehicle | |
| JP3920535B2 (en) | Vehicle detection method and vehicle detection device | |
| CN108830161B (en) | Smog identification method based on video stream data | |
| CN114639075A (en) | Method and system for identifying falling object of high altitude parabola and computer readable medium | |
| CN114841920A (en) | Flame identification method and device based on image processing and electronic equipment | |
| CN114386493A (en) | Fire detection method, system, device and medium based on flame vision virtualization | |
| CN106815567B (en) | Flame detection method and device based on video | |
| CN113256559A (en) | Rapid detection method for high-altitude object throwing | |
| CN114373162B (en) | Dangerous area personnel intrusion detection method and system for transformer substation video monitoring | |
| CN116912234B (en) | Cable stranded wire quality detection method based on image features | |
| WO2016019973A1 (en) | Method for determining stationary crowds | |
| CN113158926A (en) | High-altitude parabolic track distinguishing method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |