CN110349189A - A kind of background image update method based on continuous inter-frame difference - Google Patents

Abstract

The present invention relates to a kind of background image update method based on continuous inter-frame difference, present invention F_k(x, y) part or all of area update background image, can reduce the sensibility to Changes in weather.B is used when having foreign matter_IThe corresponding region (x, y) replaces F_k(x, y) foreground zone, the new background image corresponding region that can guarantee must be clean background.Appropriate to expand foreground zone, when can be imperfect to avoid foreign bodies detection, the prospect that can't detect be treated as F_kThe background of (x, y) and be used to update background image, cause the inaccuracy of background image.The present invention realizes the update of background image in the detection process of moving target based on background subtraction, can obtain accurate background image, and influence of the Changes in weather to moving object detection is effectively reduced.

Description

A Background Image Update Method Based on Difference Between Continuous Frames

technical field

The invention relates to a method for updating a background image when a moving target is detected by a background difference method, and more particularly relates to a method for updating a background image based on difference between consecutive frames.

Background technique

Video surveillance technology is widely used in government, finance, public security, transportation, electric power and other fields, such as the detection of track foreign object intrusion based on intelligent video surveillance in rail transit. The detection of moving objects in video surveillance, also known as foreground extraction, refers to extracting the changing area from the background image from the sequence image. Effective detection of moving targets is very important for post-processing such as target classification, tracking, and behavior understanding. The main methods of moving target detection are optical flow method, frame difference method and background difference method. The optical flow method uses the changes of pixels in the image sequence in the time domain and the correlation between adjacent frames to find the corresponding relationship between the previous frame and the current frame, and calculate the motion information of objects between adjacent frames. Noise, occluders, etc. have a great influence, and the calculation is complicated and the amount of calculation is large. The inter-frame difference method performs a difference operation on two adjacent frames of images in a video sequence, and extracts moving objects by comparing the differences between consecutive frames. This method has a simple algorithm, is insensitive to light changes, and has strong environmental adaptability, but the moving speed of the target and the time interval of the difference frame have a great influence on the detection results. The background subtraction method subtracts the current frame image from the background image, and then binarizes the resulting difference image to extract the target area different from the background image. This method is simple in principle, easy to implement and fast in detection speed, but the background image in the actual scene will change dynamically due to weather, lighting, etc., so the background image needs to be updated continuously.

These three methods have their own advantages and disadvantages, and only using one of them to detect moving targets cannot get very good detection results. The accuracy of the detection results of the background subtraction method depends on the background image. However, in many moving target detection algorithms based on the background subtraction method, the background image is immutable, or an accurate background image cannot be obtained when the background image is dynamically updated. Aiming at the problems existing in updating the background image by the background subtraction method, it is necessary to invent a method to obtain an accurate background image to improve the detection accuracy.

Contents of the invention

In order to overcome at least one defect of the above-mentioned prior art, the present invention provides a background image updating method based on difference between consecutive frames.

In order to solve the above-mentioned technical problems, the technical solution adopted in the present invention is: a method for updating background images based on the difference between consecutive frames, comprising the following steps:

S1. Collect several frames of images without foreground within a period of time in the video, sum them and take the average value as the initial background image;

S2. Read the first frame image of the video, and make a difference between the current frame image and the initial background image to obtain a differential image;

S3. Binarize the differential image to obtain a binary image, thereby separating the foreground and the background;

S4. Filter out the isolated foreground points in the binary image through the morphological opening operation, and obtain a relatively complete outline of the foreground through the closing operation;

S5. When the foreground is not separated, use the current frame image as a new background image;

S6. If the foreground is separated, then mark the minimum circumscribed rectangle of the foreground in the binary image, and appropriately expand the circumscribed rectangle; the expanded rectangular area is the foreground area, and the others are the background area; the foreground area of the current frame image uses the initial background The pixel value of the corresponding area of the image is replaced, the background area remains unchanged, and the replaced current frame image is used as the new background image;

S7. Read the next frame image of the video, make a difference between the next frame image and the new background image, and repeat steps S3 to S7.

Further, the S1 step specifically includes: selecting images of multiple frames without moving objects within a certain period of time in the video, summing them and taking an average value, and the obtained average image is used as the initial background image, namely:

In the formula, B _I (x, y) refers to the initial background image, F _i (x, y) represents the image of the i-th frame without moving objects, and N is the number of frames of images without moving objects.

Further, the formula for calculating the differential image in the step S2 is:

ΔF(x,y)=F ₁ (x,y)-B _I (x,y)

In the formula, ΔF(x, y) is the difference image, and F ₁ (x, y) is the first frame image of the video.

Further, the step S3 specifically includes: binarizing ΔF(x, y) to obtain a binary image, namely:

In the formula, B_ΔF(x, y) is a binary image, and T is the threshold value of the pixel gray value; the pixel of the binary image B_ΔF(x, y)=255 represents the foreground, and the pixel of B_ΔF(x,y)=0 The pixels represent the background.

Further, the S4 step specifically includes:

S41. Eliminate small objects through morphological opening operations; perform opening operations on binary images, and filter out isolated foreground points in images; opening operations: X ₁ is the original image, X ₂ is the image after the opening operation, and B ₁ is the structural element;

S42. Fill the small object by morphological closing operation; perform closing operation on the binary image X ₂ processed in the step S41 to obtain a foreground with complete outline; closing operation: X ₃ is the image after the closing operation, and B ₂ is the structural element.

Further, the S6 step specifically includes:

S61. Mark the circumscribed rectangles of each foreground area in the binary image B_ΔF(x, y);

S62. Appropriately expand the circumscribed rectangle, the enlarged rectangular area is regarded as the foreground area, and other areas are regarded as the background area;

S63. The current frame image F _k (x, y) corresponds to the foreground area, and replaces the foreground area of the current frame image with the area corresponding to the initial background image B _I (x, y);

S64. The area corresponding to the background of the current frame image F _k (x, y) remains unchanged,

S65. The current frame image processed in S63 and S64 is used as a new background image.

Further, the step S7 is specifically as follows: continue to read the next frame image of the video, make a difference between the next frame image and the new background image, and then repeat steps S3 to S7 to realize the moving target detection process based on the background difference method Background image updates.

Compared with the prior art, the beneficial effect is: the present invention uses F _k (x, y) to update part or all of the background image, which can reduce the sensitivity to weather changes; when there are foreign objects, use B _I (x, y) to correspond The area replaces the F _k (x, y) foreground area, which can ensure that the corresponding area of the obtained new background image must be a clean background. Appropriately expanding the foreground area can prevent the undetected foreground from being used as the background of F _k (x, y) to update the background image when foreign object detection is incomplete, resulting in inaccurate background images. The invention realizes the update of the background image in the moving target detection process based on the background difference method, can obtain accurate background images, and effectively reduces the influence of weather changes on the moving target detection.

Description of drawings

Fig. 1 is a flow chart of the method of the present invention.

Detailed ways

The accompanying drawings are for illustrative purposes only, and should not be construed as limiting the present invention; in order to better illustrate this embodiment, certain components in the accompanying drawings will be omitted, enlarged or reduced, and do not represent the size of the actual product; for those skilled in the art It is understandable that some well-known structures and descriptions thereof may be omitted in the drawings. The positional relationship described in the drawings is for illustrative purposes only, and should not be construed as limiting the present invention.

As shown in Figure 1, a background image update method based on the difference between consecutive frames includes the following steps:

Step 1. Collect several frames of images without foreground within a period of time in the video, sum them and take the average value as the initial background image;

Select multiple frames of images without moving objects within a certain period of time in the video, sum them and take the average value, and the obtained average image is used as the initial background image, namely:

In the formula, B _I (x, y) refers to the initial background image, F _i (x, y) represents the image of the i-th frame without moving objects, and N is the number of frames of images without moving objects.

Step 2. Read the first frame image of the video, and make a difference between the current frame image and the initial background image to obtain a difference image; the formula for the difference image is:

ΔF(x,y)=F ₁ (x,y)-B _I (x,y)

In the formula, ΔF(x, y) is the difference image, and F ₁ (x, y) is the first frame image of the video.

Step 3. Binarize the differential image to obtain a binary image, thereby separating the foreground and background;

First binarize ΔF(x,y) to obtain a binary image, namely:

In the formula, B_ΔF(x, y) is a binary image, and T is the threshold value of the pixel gray value; the pixel of the binary image B_ΔF(x, y)=255 represents the foreground, and the pixel of B_ΔF(x,y)=0 The pixels represent the background.

Step 4. Filter out the isolated foreground points in the binary image through the morphological opening operation, and obtain a relatively complete outline of the foreground through the closing operation;

S41. Eliminate small objects through morphological opening operations; perform opening operations on binary images, and filter out isolated foreground points in images; opening operations: X ₁ is the original image, X ₂ is the image after the opening operation, and B ₁ is the structural element;

S42. Fill the small object by morphological closing operation; perform closing operation on the binary image X ₂ processed in the step S41 to obtain a foreground with complete outline; closing operation: X ₃ is the image after the closing operation, and B ₂ is the structural element.

S5. When the foreground is not separated, use the current frame image as a new background image; namely:

B _N (x, y) = F _k (x, y)

In the formula, F _k (x, y) is the current frame image, and B _N (x, y) is the new background image.

Step 6. If the foreground is separated, then mark the minimum circumscribed rectangle of the foreground in the binary image, and appropriately expand the circumscribed rectangle; the enlarged rectangular area is the foreground area, and the others are the background area; the foreground area of the current frame image uses the initial The pixel value of the corresponding area of the background image is replaced, the background area remains unchanged, and the replaced current frame image is used as the new background image;

S61. Mark the circumscribed rectangles of each foreground area in the binary image B_ΔF(x, y);

S62. Appropriately expand the circumscribed rectangle, the enlarged rectangular area is regarded as the foreground area, and other areas are regarded as the background area;

S63. The current frame image F _k (x, y) corresponds to the foreground area, and replaces the foreground area of the current frame image with the area corresponding to the initial background image B _I (x, y);

S64. The area corresponding to the background of the current frame image F _k (x, y) remains unchanged,

S65. The current frame image F _k (x, y) processed by S63 and S64 is used as a new background image B _N (x, y).

Step 7. Read the next frame image of the video, make a difference between the next frame image and the new background image, and repeat steps S3 to S7. which is:

Continue to read the next frame image of the video, make a difference between the next frame image and the new background image, and then repeat steps S3 to S7 to realize the update of the background image during the moving target detection process based on the background difference method.

Apparently, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, rather than limiting the implementation of the present invention. For those of ordinary skill in the art, other changes or changes in different forms can be made on the basis of the above description. It is not necessary and impossible to exhaustively list all the implementation manners here. All modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (7)

1. A background image updating method based on difference between consecutive frames, is characterized in that, comprises the following steps: S1. Collect several frames of images without foreground within a period of time in the video, sum them and take the average value as the initial background image; S2. Read the first frame image of the video, and make a difference between the current frame image and the initial background image to obtain a differential image; S3. Binarize the differential image to obtain a binary image, thereby separating the foreground and the background; S4. Filter out the isolated foreground points in the binary image through the morphological opening operation, and obtain a relatively complete outline of the foreground through the closing operation; S5. When the foreground is not separated, use the current frame image as a new background image; S6. If the foreground is separated, then mark the minimum circumscribed rectangle of the foreground in the binary image, and appropriately expand the circumscribed rectangle; the expanded rectangular area is the foreground area, and the others are the background area; the foreground area of the current frame image uses the initial background The pixel value of the corresponding area of the image is replaced, the background area remains unchanged, and the replaced current frame image is used as the new background image; S7. Read the next frame image of the video, make a difference between the next frame image and the new background image, and repeat steps S3 to S7. 2. A kind of background image update method based on continuous inter-frame difference according to claim 1, it is characterized in that, described S1 step specifically comprises: select the image of multiple frames without moving target in a certain period of time of the video, and convert them After the summation, the average value is taken, and the obtained average image is used as the initial background image, that is: In the formula, B _I (x, y) refers to the initial background image, F _i (x, y) represents the i-th image without moving objects, and N is the number of frames without moving objects. 3. a kind of background image update method based on difference between continuous frames according to claim 2, is characterized in that, in the described S2 step, the formula for seeking difference image is: ΔF (x, y) = F ₁ (x, y) - B _I (x, y) In the formula, ΔF(x, y) is the difference image, and F ₁ (x, y) is the first frame image of the video. 4. A kind of background image update method based on continuous inter-frame difference according to claim 3, it is characterized in that, described S3 step specifically comprises: Binarization ΔF (x, y), obtains binary image, namely : In the formula, B_ΔF(x, y) is a binary image, and T is the threshold value of the pixel gray value; the pixel of the binary image B_ΔF(x, y)=255 represents the foreground, and the pixel of B_ΔF(x, y)=0 The pixels represent the background. 5. A kind of background image updating method based on difference between continuous frames according to claim 4, characterized in that, described S4 step specifically comprises: S41. Eliminate small objects through morphological opening operations; perform opening operations on binary images, and filter out isolated foreground points in images; opening operations: X ₁ is the original image, X ₂ is the image after the opening operation, and B ₁ is the structural element; S42. Fill the small object by morphological closing operation; perform closing operation on the binary image X ₂ processed in the step S41 to obtain a foreground with complete outline; closing operation: X ₃ is the image after the closing operation, and B ₂ is the structural element. 6. A kind of background image update method based on continuous interframe difference according to claim 5, is characterized in that, described S6 step specifically comprises: S61. Mark the circumscribed rectangles of each foreground area in the binary image B_ΔF(x, y); S62. Appropriately expand the circumscribed rectangle, the enlarged rectangular area is regarded as the foreground area, and other areas are regarded as the background area; S63. The current frame image F _k (x, y) corresponds to the foreground area, and replaces the foreground area of the current frame image with the area corresponding to the initial background image B _I (x, y); S64. The area corresponding to the background of the current frame image F _k (x, y) remains unchanged, S65. The current frame image F _k (x, y) processed in S63 and S64 is used as a new background image B _N (x, y). 7. A method for updating background images based on continuous inter-frame differences according to claim 6, wherein the step S7 is specifically: continue to read the next frame image of the video, and combine the next frame image with the new The background image is subtracted, and then steps S3 to S7 are repeated to realize updating of the background image during the moving target detection process based on the background subtraction method.