CN114187219A - Moving target shadow real-time elimination method based on red, green and blue double difference - Google Patents
Moving target shadow real-time elimination method based on red, green and blue double difference Download PDFInfo
- Publication number
- CN114187219A CN114187219A CN202111476285.2A CN202111476285A CN114187219A CN 114187219 A CN114187219 A CN 114187219A CN 202111476285 A CN202111476285 A CN 202111476285A CN 114187219 A CN114187219 A CN 114187219A
- Authority
- CN
- China
- Prior art keywords
- pixel point
- green
- red
- blue
- shadow
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 230000008030 elimination Effects 0.000 title claims description 14
- 238000003379 elimination reaction Methods 0.000 title claims description 14
- 238000004364 calculation method Methods 0.000 claims abstract description 6
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/254—Analysis of motion involving subtraction of images
-
- 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
- G06T2207/10024—Color image
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20224—Image subtraction
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention aims to provide a method for eliminating a shadow of a moving target in real time based on red, green and blue double differences, which comprises the following steps: A. shooting a background frame RGB image, and shooting a current frame RGB image containing a target; B. respectively subtracting each pixel point in the current frame RGB image from the corresponding pixel point in the background frame RGB image according to red, green and blue three channels; C. mutually subtracting the one-time difference results of the red, green and blue three-channel pixel points of each pixel point again; D. performing mean calculation on the double difference result of each pixel point according to the corresponding point to obtain the mean value of each pixel point; E. setting a threshold, comparing each pixel point with the threshold, and setting the pixel point value to be 255 if the pixel point value is larger than the threshold; and if the pixel point value is less than or equal to the threshold value, setting the pixel point value to be 0, and constructing a single-channel binary image. The method can effectively eliminate the influence of the shadow on the target and detect the pixel points of the target.
Description
Technical Field
The invention relates to the field of image processing, in particular to a moving target shadow real-time elimination method based on red, green and blue double differences.
Background
The shadow is a natural phenomenon generated by the fact that a light source is shielded by an object, and the target detection and tracking effects are seriously influenced. The main methods for eliminating the shadow include a model method and a feature method.
Modeling methods typically design models based on known shadow prior knowledge and detect and eliminate shadows accordingly. For example: zhang et al propose a robust vehicle detection method with shadow elimination, but the method for removing shadow regions by using a model proposed by the method is only suitable for relevant regions mentioned in the article, and the application range is small. Benish et al propose a technique for segmenting shadow regions using a tristimulus model (TAM) and intensity information, but this technique still requires the manual selection of shadow regions when removing shadows. Murali et al propose a technique for removing shadows from an image with uniform texture using modeling of shadow regions, but the technique is still highly dependent on the effect of the modeling. Since the shadow of the moving object itself changes as it moves, determining an effective prior is very difficult and limiting. Therefore, the shadow elimination effect of the model method on the moving target is not ideal.
The eigen method focuses on detecting and eliminating shadows using similar changes in the physical characteristics of the shadow region itself, which involve physical characteristics including brightness, color, texture, gradients, edges, etc. Generally, the physical characteristics of shadows are relatively stable, so that the characteristic method is most widely applied to the detection of the current moving objects. For example, the image log domain difference thought is utilized by the allowed macroscopics and the like to weaken the shadow area firstly and then eliminate the shadow through HSV color characteristics; the Park is to remove the shadow by using a shadow depth map and an illumination invariant feature; liyawei and the like provide a shadow weakening algorithm based on brightness and texture characteristics, and do not need early training and manual intervention; chen et al propose to detect shadows by RGB color and further detect and eliminate shadow pixels based on the distribution rule of vector distance normalized NVD. However, the physical quantities such as the brightness and the color of the image are greatly affected by the illumination and the camera imaging method, so that the stability is not high. The physical quantities such as texture, gradient, edge, etc. are highly dependent on the hyper-parameters involved in the algorithm, and thus the applicability is less than ideal. Although the latest shadow elimination technology at present sufficiently fuses shadow detection results of different physical characteristics to improve the stability of shadow elimination, the effect is still not ideal enough, and a large calculation overhead is additionally added, so that the requirements of real-time detection and elimination of the target shadow are difficult to meet.
Disclosure of Invention
The invention aims to provide a moving target shadow real-time elimination method based on red-green-blue double difference, which can effectively eliminate the influence of shadow on a target and correctly detect pixel points of the target.
The technical scheme of the invention is as follows:
the method for eliminating the shadow of the moving target in real time based on the red, green and blue double difference comprises the following steps:
A. photographing a background frame RGB image Bm×n×3Shooting a current frame RGB image F containing the targetm×n×3；
B. The current frame RGB image Fm×n×3Each pixel point in the image and the corresponding pixel point in the background frame RGB image Bm×n×3The corresponding pixel points in the middle are respectively subjected to difference according to the red, green and blue channels to obtain a double difference result delta R of the pixel points of the red channelm×nOne-time difference result delta G of green channel pixel pointsm×nOne-time difference result delta B of blue channel pixel pointsm×n；
C. One-time difference result Delta R of each pixel pointm×n、△Gm×n、△Bm×nMaking difference again to obtain the double difference result delta of red and green channels of each pixel point respectively2RGm×nDouble difference result of red and blue channels2RBm×nGreen-blue channel double difference result delta2GBm×n；
D. The double difference result delta of each pixel point2RGm×n，△2RBm×n，△2GBm×nCalculating the mean value according to the corresponding points to obtain the mean value delta of each pixel point2AVGm×n；
E. Setting threshold value, delta of each pixel point2AVGm×nRespectively comparing with a threshold value, and setting the pixel point value to be 255 if the pixel point value is larger than the threshold value; and if the pixel point value is less than or equal to the threshold value, setting the pixel point value to be 0, constructing a single-channel binary image, and obtaining the target current frame image with the shadow eliminated.
In the step B, the formula for respectively making difference of the red, green and blue three channels is as follows:
[△Rm×n,△Gm×n,△Bm×n]＝△Fm×n×3＝Bm×n×3-Fm×n×3 (1)。
in the step C, the formula of mutually subtracting the multiple difference results of the red, green and blue three-channel pixel points of the pixel point is as follows:
△2RGm×n＝△Rm×n-△Gm×n (2)
△2RBm×n＝△Rm×n-△Bm×n (3)
△2GBm×n＝△Gm×n-△Bm×n (4)。
in the step D, the mean value calculation formula is as follows:
in the above step E, the threshold value is set to 30.
The invention adopts the thought of double difference to calculate the relationship between the current frame and the background frame of the image, utilizes the characteristic that the difference values of the red, green and blue three-channel pixel points in the target shadow of the color image are similar to the difference values of the original background, respectively differentiates the three-channel pixel point difference values, then calculates the mean value of the corresponding points of the three double differences through double difference, ensures the stability of the shadow pixel points in the double difference calculation, eliminates the shadow area through a threshold value, and directly extracts the binary image of the target area.
Drawings
Fig. 1 is a background frame RGB image of embodiment 1;
FIG. 2 is an image of a target at the time of example 1, which is disturbed by shadows;
FIG. 3 is an image of a target at time two in example 1, which is disturbed by shadows;
FIG. 4 is an image of the three targets affected by shadow at time instant of example 1;
FIG. 5 is a binary diagram of the instant-target with shadow interference removed in example 1;
FIG. 6 is a binary diagram of the two targets at time instant 1 after eliminating shadow interference;
FIG. 7 is a binary diagram of the three targets eliminating the shadow interference at time instant of embodiment 1;
Detailed Description
The present invention will be described in detail with reference to the accompanying drawings and examples.
Example 1
The method for eliminating the shadow of the moving target in real time based on the red, green and blue double difference comprises the following steps:
A. photographing a background frame RGB image Bm×n×3Shooting a current frame RGB image F containing the targetm×n×3；
B. The current frame RGB image Fm×n×3Each pixel point in the image and the corresponding pixel point in the background frame RGB image Bm×n×3The corresponding pixel points in the middle are respectively subjected to difference according to the red, green and blue channels to obtain a double difference result delta R of the pixel points of the red channelm×nOne-time difference result delta G of green channel pixel pointsm×nOne-time difference result delta B of blue channel pixel pointsm×n；
The formula for making difference of red, green and blue channels is:
[△Rm×n,△Gm×n,△Bm×n]＝△Fm×n×3＝Bm×n×3-Fm×n×3 (1)；
C. one-time difference result Delta R of each pixel pointm×n、△Gm×n、△Bm×nMutually make difference again to respectively obtainRed and green channel double difference result delta to each pixel point2RGm×nDouble difference result of red and blue channels2RBm×nGreen-blue channel double difference result delta2GBm×n；
The formula for mutually subtracting the one-time difference results of the red, green and blue three-channel pixel points of the pixel point is as follows:
△2RGm×n＝△Rm×n-△Gm×n (2)
△2RBm×n＝△Rm×n-△Bm×n (3)
△2GBm×n＝△Gm×n-△Bm×n (4)；
D. the double difference result delta of each pixel point2RGm×n，△2RBm×n，△2GBm×nCalculating the mean value according to the corresponding points to obtain the mean value delta of each pixel point2AVGm×n；
E. Setting the threshold value to be 30 and delta of each pixel point2AVGm×nRespectively comparing with a threshold value, and setting the pixel point value to be 255 if the pixel point value is larger than the threshold value; and if the pixel point value is less than or equal to the threshold value, setting the pixel point value to be 0, constructing a single-channel binary image, and obtaining the target current frame image with the shadow eliminated.
The experimental results are shown in fig. 1-7, where fig. 1 is a background frame RGB image, fig. 2 is a time-one-target shadow interference image, fig. 3 is a time-two-target shadow interference image, fig. 4 is a time-three-target shadow interference image, fig. 5 is a time-one-target binary image after shadow interference is eliminated, fig. 6 is a time-two-target binary image after shadow interference is eliminated, and fig. 7 is a time-three-target binary image after shadow interference is eliminated.
The background frame and the current frame at three randomly selected moments are used as input, and the binary image of the current frame after shadow elimination can be finally obtained through calculation of the design. Fig. 5-7 represent the target binary frame data after the target is removed from the shadow interference in fig. 2-4, respectively. Obviously, the method can effectively eliminate the influence of the shadow on the target and correctly detect the pixel points of the target.
Claims (5)
1. A method for eliminating shadow of a moving target in real time based on red, green and blue double differences is characterized by comprising the following steps:
A. photographing a background frame RGB image Bm×n×3Shooting a current frame RGB image F containing the targetm×n×3；
B. The current frame RGB image Fm×n×3Each pixel point in the image and the corresponding pixel point in the background frame RGB image Bm×n×3The corresponding pixel points in the middle are respectively subjected to difference according to the red, green and blue channels to obtain a double difference result delta R of the pixel points of the red channelm×nOne-time difference result delta G of green channel pixel pointsm×nOne-time difference result delta B of blue channel pixel pointsm×n；
C. One-time difference result Delta R of each pixel pointm×n、△Gm×n、△Bm×nMaking difference again to obtain the double difference result delta of red and green channels of each pixel point respectively2RGm×nDouble difference result of red and blue channels2RBm×nGreen-blue channel double difference result delta2GBm×n；
D. The double difference result delta of each pixel point2RGm×n，△2RBm×n，△2GBm×nCalculating the mean value according to the corresponding points to obtain the mean value delta of each pixel point2AVGm×n；
E. Setting threshold value, delta of each pixel point2AVGm×nRespectively comparing with a threshold value, and setting the pixel point value to be 255 if the pixel point value is larger than the threshold value; and if the pixel point value is less than or equal to the threshold value, setting the pixel point value to be 0, constructing a single-channel binary image, and obtaining the target current frame image with the shadow eliminated.
2. The method for real-time elimination of moving object shadows based on red-green-blue double difference as claimed in claim 1, wherein:
in the step B, the formula for respectively making difference of the red, green and blue three channels is as follows:
[△Rm×n,△Gm×n,△Bm×n]＝△Fm×n×3＝Bm×n×3-Fm×n×3 (1)。
3. the method for real-time elimination of moving object shadows based on red-green-blue double difference as claimed in claim 1, wherein:
in the step C, the formula of mutually subtracting the multiple difference results of the red, green and blue three-channel pixel points of the pixel point is as follows:
△2RGm×n＝△Rm×n-△Gm×n (2)
△2RBm×n＝△Rm×n-△Bm×n (3)
△2GBm×n＝△Gm×n-△Bm×n (4)。
4. the method for real-time elimination of moving object shadows based on red-green-blue double difference as claimed in claim 1, wherein:
in the step D, the mean value calculation formula is as follows:
5. the method for real-time elimination of moving object shadows based on red-green-blue double difference as claimed in claim 1, wherein:
in the above step E, the threshold value is set to 30.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111476285.2A CN114187219B (en) | 2021-12-06 | 2021-12-06 | Moving target shadow real-time eliminating method based on red-green-blue double difference |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111476285.2A CN114187219B (en) | 2021-12-06 | 2021-12-06 | Moving target shadow real-time eliminating method based on red-green-blue double difference |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114187219A true CN114187219A (en) | 2022-03-15 |
| CN114187219B CN114187219B (en) | 2024-06-25 |
Family
ID=80603458
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111476285.2A Active CN114187219B (en) | 2021-12-06 | 2021-12-06 | Moving target shadow real-time eliminating method based on red-green-blue double difference |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114187219B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115482474A (en) * | 2022-08-24 | 2022-12-16 | 湖南科技大学 | Bridge deck vehicle load identification method and system based on high-altitude aerial image |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090087025A1 (en) * | 2007-09-29 | 2009-04-02 | Samsung Electronics Co., Ltd. | Shadow and highlight detection system and method of the same in surveillance camera and recording medium thereof |
| CN101833749A (en) * | 2009-03-12 | 2010-09-15 | 株式会社理光 | Device and method for detecting shadow in image |
| KR20100117976A (en) * | 2009-04-27 | 2010-11-04 | 삼성테크윈 주식회사 | Apparatus for estimation shadow |
| CN102147861A (en) * | 2011-05-17 | 2011-08-10 | 北京邮电大学 | Moving target detection method for carrying out Bayes judgment based on color-texture dual characteristic vectors |
| CN103164847A (en) * | 2013-04-03 | 2013-06-19 | 上海理工大学 | Method for eliminating shadow of moving target in video image |
| CN103295013A (en) * | 2013-05-13 | 2013-09-11 | 天津大学 | Pared area based single-image shadow detection method |
| CN103440628A (en) * | 2013-08-27 | 2013-12-11 | 宁波海视智能系统有限公司 | Method for removing shadow interference of moving object in video |
| CN105913441A (en) * | 2016-04-27 | 2016-08-31 | 四川大学 | Shadow removal method for improving target detection performance in video |
| CN107767390A (en) * | 2017-10-20 | 2018-03-06 | 苏州科达科技股份有限公司 | The shadow detection method and its system of monitor video image, shadow removal method |
| CN112949484A (en) * | 2021-03-01 | 2021-06-11 | 浙江大学 | High-speed railway rockfall real-time detection method based on aggregated channel features and texture features |
-
2021
- 2021-12-06 CN CN202111476285.2A patent/CN114187219B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090087025A1 (en) * | 2007-09-29 | 2009-04-02 | Samsung Electronics Co., Ltd. | Shadow and highlight detection system and method of the same in surveillance camera and recording medium thereof |
| CN101833749A (en) * | 2009-03-12 | 2010-09-15 | 株式会社理光 | Device and method for detecting shadow in image |
| KR20100117976A (en) * | 2009-04-27 | 2010-11-04 | 삼성테크윈 주식회사 | Apparatus for estimation shadow |
| CN102147861A (en) * | 2011-05-17 | 2011-08-10 | 北京邮电大学 | Moving target detection method for carrying out Bayes judgment based on color-texture dual characteristic vectors |
| CN103164847A (en) * | 2013-04-03 | 2013-06-19 | 上海理工大学 | Method for eliminating shadow of moving target in video image |
| CN103295013A (en) * | 2013-05-13 | 2013-09-11 | 天津大学 | Pared area based single-image shadow detection method |
| CN103440628A (en) * | 2013-08-27 | 2013-12-11 | 宁波海视智能系统有限公司 | Method for removing shadow interference of moving object in video |
| CN105913441A (en) * | 2016-04-27 | 2016-08-31 | 四川大学 | Shadow removal method for improving target detection performance in video |
| CN107767390A (en) * | 2017-10-20 | 2018-03-06 | 苏州科达科技股份有限公司 | The shadow detection method and its system of monitor video image, shadow removal method |
| CN112949484A (en) * | 2021-03-01 | 2021-06-11 | 浙江大学 | High-speed railway rockfall real-time detection method based on aggregated channel features and texture features |
Non-Patent Citations (5)
| Title |
|---|
| 李萌;郑娟毅;门瑜;: "基于帧差法的运动车辆阴影去除算法", 电视技术, no. 10, 17 October 2016 (2016-10-17) * |
| 王静静;林明秀;魏颖;: "基于灰度相关的帧间差分和背景差分相融合的实时目标检测", 中南大学学报(自然科学版), no. 1, 10 September 2009 (2009-09-10) * |
| 管业鹏;顾伟康;: "二维场景阴影区域的自动鲁棒分割", 电子学报, no. 04, 25 April 2006 (2006-04-25) * |
| 胡金杰;王远弟;: "视频监控系统中运动目标的阴影消除", 应用数学与计算数学学报, no. 01, 15 June 2010 (2010-06-15) * |
| 陈前: "复杂背景下运动目标检测和阴影消除算法研究", 《中国优秀硕士学位论文全文数据库》, 15 February 2020 (2020-02-15) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115482474A (en) * | 2022-08-24 | 2022-12-16 | 湖南科技大学 | Bridge deck vehicle load identification method and system based on high-altitude aerial image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114187219B (en) | 2024-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101518722B1 (en) | Image enhancement processing method and apparatus for distortion correction by air particle like fog | |
| CN108549874B (en) | Target detection method, target detection equipment and computer-readable storage medium | |
| CN102385753B (en) | Illumination-classification-based adaptive image segmentation method | |
| CN107016691B (en) | Moving target detecting method based on super-pixel feature | |
| WO2015180527A1 (en) | Image saliency detection method | |
| US8189913B2 (en) | Method for detecting shadow of object | |
| CN110570360B (en) | Retinex-based robust and comprehensive low-quality illumination image enhancement method | |
| KR101051459B1 (en) | Apparatus and method for extracting edges of an image | |
| CN107256539B (en) | Image sharpening method based on local contrast | |
| CN106600569B (en) | Signal lamp color effect enhancement processing method and device | |
| CN103065334A (en) | Color cast detection and correction method and device based on HSV (Hue, Saturation, Value) color space | |
| CN104504722A (en) | Method for correcting image colors through gray points | |
| CN107481214A (en) | A kind of twilight image and infrared image fusion method | |
| CN103942812A (en) | Moving object detection method based on Gaussian mixture and edge detection | |
| CN114187219A (en) | Moving target shadow real-time elimination method based on red, green and blue double difference | |
| CN110188640B (en) | Face recognition method, face recognition device, server and computer readable medium | |
| Tan et al. | Image haze removal based on superpixels and Markov random field | |
| JP2017073639A5 (en) | ||
| Shiting et al. | Clustering-based shadow edge detection in a single color image | |
| CN108010050B (en) | Foreground detection method based on adaptive background updating and selective background updating | |
| CN113379631B (en) | Image defogging method and device | |
| CN113850878B (en) | Color histogram-based non-light source AWB (AWB) estimation method and system | |
| Li et al. | A perceptual color edge detection algorithm | |
| CN108389219B (en) | Weak and small target tracking loss re-detection method based on multi-peak judgment | |
| CN109191481B (en) | Rapid threshold segmentation method for shoe print image under poor exposure condition |
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 |