CN109685827A - A kind of object detecting and tracking method based on DSP - Google Patents

A kind of object detecting and tracking method based on DSP Download PDF

Info

Publication number
CN109685827A
CN109685827A CN201811453586.1A CN201811453586A CN109685827A CN 109685827 A CN109685827 A CN 109685827A CN 201811453586 A CN201811453586 A CN 201811453586A CN 109685827 A CN109685827 A CN 109685827A
Authority
CN
China
Prior art keywords
target
pixel
image
frame image
value
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.)
Granted
Application number
CN201811453586.1A
Other languages
Chinese (zh)
Other versions
CN109685827B (en
Inventor
邹卫军
凌永鹏
翟弘绅
单崇铭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Science and Technology
Original Assignee
Nanjing University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University of Science and Technology filed Critical Nanjing University of Science and Technology
Priority to CN201811453586.1A priority Critical patent/CN109685827B/en
Publication of CN109685827A publication Critical patent/CN109685827A/en
Application granted granted Critical
Publication of CN109685827B publication Critical patent/CN109685827B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G06T7/251—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • 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
    • G06T2207/20024—Filtering details

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Databases & Information Systems (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Image Analysis (AREA)

Abstract

The object detecting and tracking method based on DSP that the invention discloses a kind of, comprising the following steps: the size of initialization wave door and its surrounding background first, and the position of wave door in the picture is set makes that it includes targets to be tracked；Binaryzation and filtering processing are carried out to Bo Mennei image later；Then it marks connected domain and determines target template；Further according to the position of target in the position acquisition a later frame image of target in two continuous frames image；The related coefficient in a later frame image between target and target template is finally sought, and judges that being to continue with tracking is also off tracking.Object detecting and tracking method computational complexity of the invention is low, and is exaggerated the correlation of target template with target in image, ensure that the stability of image trace, tracking accuracy is high, and real-time is good, is able to satisfy requirement of the image tracking system to tracking.

Description

A kind of object detecting and tracking method based on DSP
Technical field
The invention belongs to view synthesis and tracking control technology, especially a kind of target detection based on DSP with Track method.
Background technique
In every application of Digital Video Processing and computer vision field, target detection and tracking are that a weight is basic And important task.Some preferable fields of development prospect, robot control, it is based drive identification, view-based access control model control, Augmented reality, video scene monitoring, navigational guidance require to use object detecting and tracking technology.In computer vision field, The detection and tracking of target remain the higher research field of temperature.The emphasis of research is also by pursuing merely high-precision and high-stability Simulation analysis, the industrial application direction small to real-time operand develop, how as far as possible meet performance indicator before It puts, reduces the difficult point that the requirement to hardware is realistic objective detection and follow-up study direction.In order to reduce the requirement to hardware, Chinese patent CN201010121006.6 proposes a kind of object detecting and tracking method and Digital Image Processing based on DSP System, but this method operand is big, versatility and real-time are poor, are being now the video of main hardware platform with multi-core DSP It is difficult to be well used in image processing system.Study the target inspection that a kind of precision is high, computational complexity is low, practicability is good Survey has great importance with tracking.
Summary of the invention
Technical problem solved by the invention is to provide that a kind of operand is low, be easily achieved and is able to satisfy high real-time system The object detecting and tracking method that system requires.
The technical solution for realizing the aim of the invention is as follows: a kind of object detecting and tracking method based on DSP, including with Lower step:
First frame image in step 1, acquisition video；
Step 2, the size according to the first frame image initial wave door of acquisition, initialize wave according to the size of wave door later The size of door surrounding background, and position of the wave door in first frame image is set makes it includes target to be tracked, count wave door and The grey level histogram of wave door surrounding background；
Step 3, the threshold value that binaryzation is determined according to the grey level histogram, and Bo Mennei image is carried out according to the threshold value Binary conversion treatment；
Step 4 is filtered the Bo Mennei image after binary conversion treatment；
Step 5, using each region of the filtered Bo Mennei image of neighbourhood signatures' algorithm tag, obtain several connections Region；
The size of all connected regions in step 6, statistic procedure 5, and descending arrangement is carried out to connected region size, it obtains The wherein minimum circumscribed rectangle in largest connected region, and as target template；
Step 7 obtains target position in the center, that is, first frame image in the largest connected region, and basis should later Center obtains target position in the second frame image, and updates target template；
Step 8 obtains target position in a later frame image according to target position in two continuous frames image；
Step 9 seeks the target in a later frame image and the related coefficient between target template, and according to phase relation Number judgement be repeat step 8 continue tracking be also off tracking；Step 8 is repeated if related coefficient Q is more than or equal to preset threshold p Continue to track；Otherwise frequency of failure n is incremented by 1, and judges that n and the relationship of frequency of failure preset threshold q stop if n >=q Tracking, on the contrary expand the size of wave door as unit of Pixel-level, it repeats step 8 and continues to track.
Compared with prior art, the present invention its remarkable advantage: 1) present invention improves by using diamond search algorithm With efficiency, operand can be reduced under the premise of guaranteeing search effect；2) taking the weight of target template update in the present invention Value is associated with related coefficient, is avoided and is updated error message into target template when tracking effect is bad, ensure that subsequent The stability of image trace；3) by going averaging method to seek related coefficient in the present invention, while reducing operand, amplify mesh Mark the correlation of template and target in image.
The present invention is described in further detail with reference to the accompanying drawing.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow charts of the object detecting and tracking method of DSP.
Fig. 2 is tracking object delineation to be detected in a certain frame image in the embodiment of the present invention.
Fig. 3 is using search pattern schematic diagram used in diamond search algorithm in the embodiment of the present invention, wherein figure (a) is 5 × 5 diamond search template, the diamond search template that figure (b) is 3 × 3.
Fig. 4 is target following result schematic diagram in a certain frame image in the embodiment of the present invention.
Specific embodiment
In conjunction with Fig. 1, a kind of object detecting and tracking method based on DSP of the present invention, comprising the following steps:
First frame image in step 1, acquisition video.
Step 2, the size according to the first frame image initial wave door of acquisition, initialize wave according to the size of wave door later The size of door surrounding background, and position of the wave door in first frame image is set makes it includes target to be tracked, count wave door and The grey level histogram of wave door surrounding background.
Further, according to the size of the first frame image initial wave door of acquisition, specifically: according to first frame image The size of the size initialization wave door of target to be tracked in size and image, it is assumed that the size of the image of acquisition is w × h, to The size for tracking target minimum circumscribed rectangle is x × y, and wave door size is w' × h', wherein x < w'≤w, y < h'≤h；
Further, wave door surrounding background size is determined according to the size of wave door, specifically: if by the border extension of wave door Dry pixel.
Step 3, the threshold value that binaryzation is determined according to grey level histogram, and two-value is carried out to Bo Mennei image according to the threshold value Change processing.
Further, the threshold value of binaryzation is determined according to grey level histogram, specifically: will in grey level histogram represent to Threshold value of the corresponding gray value of low ebb as binaryzation between track target and two wave crests of its background.
Step 4 is filtered the Bo Mennei image after binaryzation.
Further, filtering processing is specifically handled using opening operation.
Step 5, using each region of the filtered Bo Mennei image of neighbourhood signatures' algorithm tag, obtain several connections Region.
Further, using each region of the image of the filtered Bo Mennei of neighbourhood signatures' algorithm tag, wherein neighborhood Labeling algorithm specifically:
To filtered Bo Mennei image by from left to right, from top to bottom in a manner of scanned pixel-by-pixel, until scanning The all pixels of complete Bo Mennei image are directly moved to next pixel if the pixel value of Current Scan is 0；If currently sweeping The pixel value retouched be 1, then be marked according to the left pixel of current pixel, upside pixel, wherein the size of mark value with New connected domain increases, and specifically includes following 4 kinds of situations:
(1) pixel value of left side and upside is 0, indicates that the pixel of Current Scan is the boundary of a new connected domain, Then assign one new mark value of pixel of Current Scan；
(2) pixel value is only existed in the pixel value of left side and upside is 1, then assigns the pixel and pixel of Current Scan The identical mark value of element marking value that value is 1；
(3) left side is 1 with the pixel value of upside and mark value is identical, then assigns the pixel and mark value phase of Current Scan Same mark value；
(4) left side is 1 with the pixel value of upside but mark value is different, then the mark value for assigning the pixel of Current Scan is The smallest mark value in mark value.
The size of all connected regions in step 6, statistic procedure 5, and descending arrangement is carried out to connected region size, it obtains The wherein minimum circumscribed rectangle in largest connected region, and as target template.
Step 7 obtains target position in the center, that is, first frame image in largest connected region, later according to the center Target position in the second frame image is obtained, and updates target template.
Further, target position in the second frame image is obtained according to the center in largest connected region, specifically:
Step 7-1, the center of wave door is moved to the center in largest connected region；
Step 7-2, Bo Mennei image match with target template by template matching algorithm and obtain the second frame image Middle target position.
Further, template matching algorithm specifically uses diamond search algorithm in step 7-2.
Further, target template is updated, specifically:
Assuming that TkFor the target template for carrying out template matching for kth frame image, Tk+1To be used for+1 frame figure of kth after update Target template as carrying out template matching, updates the formula of target template are as follows:
In formula, MkFor in kth frame image by template matching algorithm obtain centered on target position, with template size For the target template T of coverage areakBest match, α be update weight, cmaxFor the related coefficient of template matching output, τtFor The threshold value whether target template of setting updates, by formula it is found that working as related coefficient cmaxGreater than threshold tautWhen, to target template It is updated, otherwise remains unchanged.
Step 8 obtains target position in a later frame image according to target position in two continuous frames image.
Further, target position in a later frame image is obtained according to target position in two continuous frames image, Specifically:
Step 8-1, using target position in Least Square in Processing two continuous frames image, a later frame figure is thus obtained The region locating for target as in；
Step 8-2, Jiang Bomen is moved to the center in region, carries out template to Bo Mennei image and updated target template Matching obtains target position in current frame image.
Step 9 is sought the target in a later frame image and the related coefficient between target template, and is sentenced according to related coefficient Disconnected is to repeat step 8 and continue tracking to be also off tracking.
Further, step 9 seeks the target in a later frame image and the related coefficient between target template, and according to phase Relationship number judgement be repeat step 8 continue tracking be also off tracking, specifically:
Step 9-1, it in step 8 in a later frame image centered on target position, intercepts identical as target template big Small region is as the target in a later frame image；
Step 9-2, the target in a later frame image and the related coefficient Q between target template, formula used are sought are as follows:
In formula, xiFor the gray value of ith pixel in target template,For the pixel grey scale mean value of target template, yiFor figure The gray value of ith pixel in target as in,For the pixel grey scale mean value of the target in image；
Step 9-3, judge the relationship of related coefficient Q Yu preset threshold p, if Q >=p, repeatedly step 8 continues to track；Instead Frequency of failure n is incremented by 1, and execute step 9-4；Wherein, the initial value of frequency of failure n is 0；
Step 9-4, the relationship for judging n Yu frequency of failure preset threshold q stops tracking if n >=q, otherwise with Pixel-level Expand the size of wave door for unit, repeats step 8 and continue to track.
Embodiment
In conjunction with Fig. 1, the present invention is based on the object detecting and tracking methods of DSP, including the following contents:
1, wave door size and Bo Men surrounding background size are initialized
Depending on size of the size of wave door by the image size and target that acquire, major requirement is that wave door being capable of complete terrestrial reference The position of target out.The a certain frame image acquired in the present embodiment is as shown in Fig. 2, size is 1600 pixel *, 900 pixel, image Middle target sizes are 80 pixel *, 60 pixel, and Jiang Bomen is dimensioned to 120 pixel *, 120 pixel, and wave door surrounding background size is set It is set to 120 pixel *, 30 pixel.
2, binarization threshold is determined
The grey level histogram for counting Bo Men and Bo Men surrounding background obtains the histogram with bimodal distribution, chooses wave door Between threshold value of the corresponding gray value 20 of low ebb as binaryzation.
3, binaryzation Bo Mennei image
Following binary conversion treatment is carried out to Bo Mennei image according to determining binarization threshold:
Wherein, f (x, y) is the original gray value of Bo Mennei pixel, fT(x, y) is the gray scale for handling pixel in late pulse Value.
4, the Bo Mennei image of binaryzation is filtered
Handled using opening operation, specially first use size for 3 × 3 each of structural element scan image picture Element, the pixel phase "AND" covered with each of structural element pixel with it, if all be 0, the pixel be 0, otherwise for 1.Use again size for 3 × 3 each of structural element scan image pixel, with each of structural element pixel with Its pixel phase "AND" covered, if being all 1, otherwise it is 0 which, which is 1,.
5, using each region of the image of the filtered Bo Mennei of neighbourhood signatures' algorithm tag, several connected regions are obtained Domain；
6, the size of all connected regions of statistics label, and descending arrangement is carried out to connected region size, it obtains wherein The boundary rectangle in largest connected region, and as target template；
7, target position in the center, that is, first frame image in largest connected region is obtained, using diamond search algorithm root Target position in the second frame image is obtained according to the center.Concrete operations are as follows:
It is output with template image and target image, the best region where target is first found with big diamond search algorithm, Using this position as the center of small diamond search, then find with small diamond search algorithm the exact position of target.Big diamond shape The search pattern that searching algorithm uses is the template of 5 × 5 sizes, and the search pattern that small diamond search algorithm uses is 3 × 3 sizes Template, big diamond search template and small diamond search template are as shown in Figure 3.Diamond search algorithm Reusability diamond search mould Plate scans for, until that the smallest point of this search error appears in the center of template, obtained search error is the smallest Point is optimal match point.
8, target template update method
Wherein TkFor the target template for carrying out template matching for kth frame image, Tk+1To be used for+1 frame figure of kth after update Target template as carrying out template matching, MkWith target position to be by what template matching algorithm obtained in kth frame image The heart, using template size as the target template T of coverage areakBest match.α is to update weight, cmaxFor template matching output Related coefficient, α=0.1c in the present embodimentmax, τtFor the threshold value whether target template of setting updates, τ in the present embodimentt= 0.75, as related coefficient cmaxGreater than threshold tautWhen, target template is updated, is otherwise remained unchanged.
9, target position in a later frame image is obtained according to target position in two continuous frames image；
Using the position where target in front cross frame image, target institute in a later frame image is extrapolated using least square method The region at place.Least square method formula is as follows:
Y=ax+b
Wherein a be matched curve slope, b be matched curve intercept, x be target in the horizontal direction pixel seat Mark, y for target pixel coordinate in vertical direction,For front cross frame target pixel coordinate in the horizontal direction it is equal Value,For target pixel coordinate mean value in vertical direction.
10, Jiang Bomen is moved to the center in region locating for target, to the image and the updated target of previous frame of Bo Mennei Template carries out template matching, obtains target position in current frame image.
11, the target in image and the related coefficient between target template, formula used are calculated are as follows:
In formula, xiFor the gray value of ith pixel in target template,For the pixel grey scale mean value of target template, yiFor figure The gray value of ith pixel in target as in,For the pixel grey scale mean value of the target in image；Related coefficient Q ∈ [- 1, 1], and Q is bigger, and correlation is higher.
12, judge the relationship of related coefficient Yu set threshold value
0.75 is set by related coefficient preset threshold p in the present embodiment, sets 50 for frequency of failure preset threshold q, Judge the relationship of related coefficient Q Yu preset threshold p, if Q >=p, empty fail count, continue to track, target following result is such as Shown in Fig. 4；If Q < p, fail count n is incremented by 1, judges that n and the relationship of frequency of failure preset threshold q stop if n >=q It only tracks, if n < q, wave door is expanded as into 180 pixel *, 180 pixel, continues to track.
Object detecting and tracking method based on DSP of the invention, computational complexity is low, and is exaggerated target template and figure The correlation of target as in ensure that the stability of image trace, and tracking accuracy is high, and real-time is good, is able to satisfy image trace system The requirement united to tracking.

Claims (10)

1. a kind of object detecting and tracking method based on DSP, which comprises the following steps:
First frame image in step 1, acquisition video；
Step 2, the size according to the first frame image initial wave door of acquisition, initialize wave Men Si according to the size of wave door later The size of all backgrounds, and position of the wave door in first frame image is set makes to count Bo Men and Bo Men it includes target to be tracked The grey level histogram of surrounding background；
Step 3, the threshold value that binaryzation is determined according to the grey level histogram, and two-value is carried out to Bo Mennei image according to the threshold value Change processing；
Step 4 is filtered the Bo Mennei image after binary conversion treatment；
Step 5, using each region of the filtered Bo Mennei image of neighbourhood signatures' algorithm tag, obtain several connected regions Domain；
The size of all connected regions in step 6, statistic procedure 5, and descending arrangement is carried out to connected region size, it obtains wherein The minimum circumscribed rectangle in largest connected region, and as target template；
Step 7 obtains target position in the center, that is, first frame image in the largest connected region, later according to the center Target position in the second frame image is obtained, and updates target template；
Step 8 obtains target position in a later frame image according to target position in two continuous frames image；
Step 9 is sought the target in a later frame image and the related coefficient between target template, and is sentenced according to related coefficient Disconnected is to repeat step 8 and continue tracking to be also off tracking；Step 8 is repeated if related coefficient Q is more than or equal to preset threshold p to continue Tracking；Otherwise frequency of failure n is incremented by 1, and judges the relationship of n Yu frequency of failure preset threshold q, if n >=q, stops tracking, Otherwise expands the size of wave door as unit of Pixel-level, repeat step 8 and continue to track.
2. the object detecting and tracking method according to claim 1 based on DSP, which is characterized in that in step 1:
The size of the first frame image initial wave door according to acquisition, specifically: according to the size of first frame image and The size of the size initialization wave door of target to be tracked in image,
Assuming that the size of the image of acquisition is w × h, the size of target minimum circumscribed rectangle to be tracked is x × y, and wave door size is W' × h', wherein x < w'≤w, y < h'≤h；
The size according to wave door determines wave door surrounding background size, specifically: by several pixels of the border extension of wave door.
3. the object detecting and tracking method according to claim 1 or 2 based on DSP, which is characterized in that described in step 3 The threshold value of binaryzation is determined according to grey level histogram, specifically: target to be tracked and its background will be represented in grey level histogram Two wave crests between threshold value of the corresponding gray value of low ebb as binaryzation.
4. the object detecting and tracking method according to claim 3 based on DSP, which is characterized in that filtered described in step 4 Wave processing is specifically handled using opening operation.
5. the object detecting and tracking method according to claim 1 based on DSP, which is characterized in that utilized described in step 5 Each region of the image of the filtered Bo Mennei of neighbourhood signatures' algorithm tag, wherein neighbourhood signatures' algorithm specifically:
To filtered Bo Mennei image by from left to right, from top to bottom in a manner of scanned pixel-by-pixel, until scan through wave The all pixels of image in door are directly moved to next pixel if the pixel value of Current Scan is 0；If Current Scan Pixel value is 1, then is marked according to the left pixel of current pixel, upside pixel, wherein the size of mark value is with new Connected domain increases, and specifically includes following 4 kinds of situations:
(1) pixel value of left side and upside is 0, indicates that the pixel of Current Scan is the boundary of a new connected domain, then assigns Give one new mark value of pixel of Current Scan；
(2) pixel value is only existed in the pixel value of left side and upside is 1, then assigns the pixel and the pixel of Current Scan The identical mark value of element marking value that value is 1；
(3) left side is 1 with the pixel value of upside and mark value is identical, then assigns the pixel and the mark value phase of Current Scan Same mark value；
(4) left side is 1 with the pixel value of upside but mark value is different, then it is described for assigning the mark value of the pixel of Current Scan The smallest mark value in mark value.
6. the object detecting and tracking method according to claim 1 based on DSP, which is characterized in that basis described in step 7 The center in largest connected region obtains target position in the second frame image, specifically:
Step 7-1, the center of wave door is moved to the center in the largest connected region；
Step 7-2, Bo Mennei image match with the target template by template matching algorithm and obtain the second frame image Middle target position.
7. the object detecting and tracking method according to claim 6 based on DSP, which is characterized in that template in step 7-2 Matching algorithm specifically uses diamond search algorithm.
8. the object detecting and tracking method according to claim 7 based on DSP, which is characterized in that updated described in step 7 Target template, specifically:
Assuming that TkFor the target template for carrying out template matching for kth frame image, Tk+1To be carried out after update for+1 frame image of kth The target template of template matching updates the formula of target template are as follows:
In formula, MkFor in kth frame image by template matching algorithm obtain centered on target position, with template size be cover The target template T of lid rangekBest match, α be update weight, cmaxFor the related coefficient of template matching output, τtFor setting The threshold value that whether updates of target template, by formula it is found that working as related coefficient cmaxGreater than threshold tautWhen, target template is carried out It updates, otherwise remains unchanged.
9. the object detecting and tracking method according to claim 1 based on DSP, which is characterized in that basis described in step 8 Target position obtains target position in a later frame image in two continuous frames image, specifically:
Step 8-1, it using target position in Least Square in Processing two continuous frames image, thus obtains in a later frame image Region locating for target；
Step 8-2, Jiang Bomen is moved to the center in the region, carries out to Bo Mennei image and the updated target template Template matching obtains target position in current frame image.
10. the object detecting and tracking method according to claim 1 based on DSP, which is characterized in that asked described in step 9 Take the target in a later frame image and the related coefficient between target template, and according to related coefficient judgement be repeat step 8 after Continuous tracking is also off tracking, specifically:
Step 9-1, it centered on target position in a later frame image described in step 8, intercepts identical as target template big Small region is as the target in a later frame image；
Step 9-2, the target in a later frame image and the related coefficient Q between target template, formula used are sought are as follows:
In formula, xiFor the gray value of ith pixel in target template,For the pixel grey scale mean value of target template, yiFor in image Target in ith pixel gray value,For the pixel grey scale mean value of the target in image；
Step 9-3, the relationship for judging the related coefficient Q and preset threshold p, if Q >=p, repeatedly step 8 continues to track；Instead Frequency of failure n is incremented by 1, and execute step 9-4；Wherein, the initial value of frequency of failure n is 0；
Step 9-4, the relationship for judging n Yu frequency of failure preset threshold q stops tracking if n >=q, otherwise is single with Pixel-level Position expands the size of wave door, repeats step 8 and continues to track.
CN201811453586.1A 2018-11-30 2018-11-30 Target detection and tracking method based on DSP Active CN109685827B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811453586.1A CN109685827B (en) 2018-11-30 2018-11-30 Target detection and tracking method based on DSP

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811453586.1A CN109685827B (en) 2018-11-30 2018-11-30 Target detection and tracking method based on DSP

Publications (2)

Publication Number Publication Date
CN109685827A true CN109685827A (en) 2019-04-26
CN109685827B CN109685827B (en) 2022-09-06

Family

ID=66185165

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811453586.1A Active CN109685827B (en) 2018-11-30 2018-11-30 Target detection and tracking method based on DSP

Country Status (1)

Country Link
CN (1) CN109685827B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110543809A (en) * 2019-06-24 2019-12-06 台州宏达电力建设有限公司 Electric power operation site risk supervision intrusion video identification method
CN110930428A (en) * 2020-02-19 2020-03-27 成都纵横大鹏无人机科技有限公司 Target tracking method and device, electronic equipment and storage medium
CN111145218A (en) * 2019-12-30 2020-05-12 华南理工大学 Mini-LED chip precision positioning method based on YOLO algorithm
CN112199972A (en) * 2020-10-28 2021-01-08 普联技术有限公司 Method for identifying positioning point
WO2024065389A1 (en) * 2022-09-29 2024-04-04 京东方科技集团股份有限公司 Method and system for detecting camera interference, and electronic device

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831617A (en) * 2012-07-17 2012-12-19 聊城大学 Method and system for detecting and tracking moving object
CN106296725A (en) * 2015-06-12 2017-01-04 富泰华工业（深圳）有限公司 Moving target detects and tracking and object detecting device in real time

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831617A (en) * 2012-07-17 2012-12-19 聊城大学 Method and system for detecting and tracking moving object
CN106296725A (en) * 2015-06-12 2017-01-04 富泰华工业（深圳）有限公司 Moving target detects and tracking and object detecting device in real time

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110543809A (en) * 2019-06-24 2019-12-06 台州宏达电力建设有限公司 Electric power operation site risk supervision intrusion video identification method
CN111145218A (en) * 2019-12-30 2020-05-12 华南理工大学 Mini-LED chip precision positioning method based on YOLO algorithm
CN111145218B (en) * 2019-12-30 2023-04-07 华南理工大学 Mini-LED chip precision positioning method based on YOLO algorithm
CN110930428A (en) * 2020-02-19 2020-03-27 成都纵横大鹏无人机科技有限公司 Target tracking method and device, electronic equipment and storage medium
CN112199972A (en) * 2020-10-28 2021-01-08 普联技术有限公司 Method for identifying positioning point
WO2024065389A1 (en) * 2022-09-29 2024-04-04 京东方科技集团股份有限公司 Method and system for detecting camera interference, and electronic device

Also Published As

Publication number Publication date
CN109685827B (en) 2022-09-06

Similar Documents

Publication Publication Date Title
CN109685827A (en) A kind of object detecting and tracking method based on DSP
JP2915894B2 (en) Target tracking method and device
CN111563442A (en) Slam method and system for fusing point cloud and camera image data based on laser radar
CN107657639A (en) A kind of method and apparatus of quickly positioning target
CN108322724B (en) Image solid matching method and binocular vision equipment
CN104794421A (en) QR (quick response) code positioning and recognizing methods
CN109919002B (en) Yellow stop line identification method and device, computer equipment and storage medium
CN107480603B (en) Synchronous mapping and object segmentation method based on SLAM and depth camera
US20060050788A1 (en) Method and device for computer-aided motion estimation
CN105488811A (en) Depth gradient-based target tracking method and system
CN111415374A (en) KVM system and method for monitoring and managing scenic spot pedestrian flow
JP3490482B2 (en) Edge and contour extraction device
CN108022245B (en) Facial line primitive association model-based photovoltaic panel template automatic generation method
CN109146918A (en) A kind of adaptive related objective localization method based on piecemeal
TWI383690B (en) Method for image processing
CN109658441A (en) Foreground detection method and device based on depth information
Cerri et al. Free space detection on highways using time correlation between stabilized sub-pixel precision IPM images
CN112528994B (en) Free angle license plate detection method, license plate recognition method and recognition system
CN103778625A (en) Surface feature intelligent searching technique based on remote sensing image variation detecting algorithm
CN107274382B (en) State identification method and device of hard pressing plate and electronic equipment
CN108765444A (en) Ground T shape Moving objects detection and location methods based on monocular vision
CN113205494A (en) Infrared small target detection method and system based on adaptive scale image block weighting difference measurement
CN111161219B (en) Robust monocular vision SLAM method suitable for shadow environment
CN108074264A (en) A kind of classification multi-vision visual localization method, system and device
KR101284252B1 (en) Curvature Field-based Corner Detection

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant