KR100332639B1 - Moving object detection method using a line matching technique - Google Patents

Abstract

The present invention relates to a method for detecting a moving object in real time in a camera moving with pan / tilt. After detecting the camera pan / tilt by each line rather than a motion search for the entire area of the image, By using this, the edge of the image can be detected and the coordinates of the moving object can be used to reduce the amount of computation and improve the speed of motion detection, and the moving part can be accurately detected by using the edge portion of the image without using the entire area of the image. have.

Description

Moving object detection method using line matching technique {MOVING OBJECT DETECTION METHOD USING A LINE MATCHING TECHNIQUE}

The present invention relates to a method of detecting a moving object, and more particularly, to a method of detecting a moving object in real time using a line matching technique in a camera moving with pan / tilt.

In general, a security system using a camera is a system that monitors a specific area using a fixed camera. Motion detection methods for monitoring a specific area using a camera include a method of detecting 8 parameters using a previous image and a current image and a method of using block matching. In the method using 8 parameters, for each position of an image, a weight least mean square error (WLMS) is obtained using an unknown 8 parameter related to a relationship between two images, and then a focal length and rotation angle satisfying the same are calculated and calculated. It is a method of detecting moving objects by inverting an image using a focal length and a rotation angle. On the other hand, a method using block matching is a method of detecting a motion of a center portion of each divided block after segmenting an image using block matching and predicting the movement of a moving object using the detected nine motions.

The conventional camera motion detection method increases the computational amount by inversely converting the focal length and the rotation angle obtained by the WLMS operation on the entire image area, which requires a lot of time to detect the moving object in the image data obtained while the camera is moving. In addition, the moving object motion detection method by block matching has a problem that the motion of the moving object cannot be accurately calculated by the influence of the noise when the noise or the moving object exists in the input image of the selected block. Occurs. In addition, a problem arises in that an error of an operation value for calculating a motion of a moving object with respect to image data input in real time increases.

Accordingly, the present invention is to solve such a problem, the present invention provides a real-time detection method of a moving object that can reduce the amount of calculation according to the detection of the moving object and improve the calculation speed by using a line matching algorithm. There is this.

Another object of the present invention is to provide a real-time detection method of a moving object which is insensitive to noise on an input image of a camera and is not affected by a moving object of the input image.

1 is a diagram illustrating a security system employing a moving object detection method using a line matching technique according to the present invention.

2 is a flowchart illustrating a moving object detection method according to the present invention.

6 is a view for explaining a motion detection algorithm of the entire image of the moving object detection method according to the present invention.

<Explanation of symbols for main parts of the drawings>

100: camera 110: pan / tilt driver

120: computer system

In order to achieve the above object, a feature of the present invention is a method of detecting a moving object in a camera moving with a pan / tilt driver, comprising: obtaining a plurality of video images using the moving camera; Extracting edge components of each of the plurality of video images to detect the moving object within the plurality of video images; Comparing the edge components extracted from each of the plurality of video images with each other; Calculating coordinates of the moving object by comparing the edge components; And adjusting the movement of the pan / tilt driver using the calculated coordinates.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram for implementing a real-time detection method of a moving object using a line matching technique according to the present invention.

In FIG. 1, a security system employing a real-time detection method of a moving object using the line matching technique of the present invention includes a camera 100, a pan / tilt driver 110, and a computer system 120. The moving object detection method of the present invention will be described in detail with reference to the security system exemplarily shown in FIG.

The image input through the camera 100 is analyzed and detected by the computer system 120. The computer system 120 calculates the coordinates of the image target to control the pan / tilt driver 110 to provide an automatic tracking and monitoring function. Perform. Computer system 120 and pan / tilt driver 110 are interfaced by RS-232C.

2 is a flowchart illustrating a real-time detection method of a moving object using a line matching technique according to the present invention. Hereinafter, a moving object detection method using the line matching method of the present invention will be described in detail with reference to FIG. 2 and FIG. 1. In FIG. 2, an image is input through the camera 100 (step S10), In order to obtain the motion of a moving object between the two video images for the current video image and the previous video image with respect to the two images input through the camera 100, the lines are not compared with the entire video image, but each line at the same position of the two video images While moving from side to side, the difference between the two image images is recorded in the memory of the computer system 120 to perform image analysis and detection (step S20). At this time, it is determined whether or not to track the moving object in the video image without directly using the data of the video image input in step S10 (step S30). According to the present invention, whether the moving object is tracked in the video image may be determined depending on whether elements such as the size and color of the moving object meet a predetermined condition. If the determination result in step S30 is negative, the process returns to step S10 and repeats the processing in steps S10 to S30. If the determination in step S30 is positive, the following equation (1) is used. The edge operation of the video image is performed by the computer system 120, and then the coordinates of the moving object of the video image are calculated using the edge operation (S40). The camera 100 moves left and right and up and down to compare the image image lines, and stores the difference value in the memory of the computer system 120. The computer system 120 stores the smallest value among the stored difference values. Think of it as a movement. As described above, the reason for calculating the difference between the two image images by using the edge of the image image after calculating the edge of the image image is that if the difference between the two image images is obtained as the edge component of the image image, the two image images do not match. This is because the difference is large.

I in Equation 1 is a position of a pixel to be shifted, n denotes an image size, I1 denotes a current image frame, I2 denotes a previous image frame, and Di denotes when the current image frame is shifted i times. It represents the sum of difference values of all pixels for any line. In Equation 1, the pan value is exemplarily denoted by x. However, if y is used instead of x, the tilt value can be obtained.

On the other hand, since the current video frame and the next frame are input at high speed intervals, since the difference in movement in the input video image is small, the calculation time is limited by moving the line to the left and right and limiting the comparison range. You can get it. In addition, the operation time may be further reduced by using the movement of the pan / tilt from the pan / tilt driver 110. For example, if the pan / tilt driver 110 knows that the pan has moved 10 to the right while changing from the previous video frame to the current video frame, the pan / tilt driver 110 may move the video image by 10 and use the 3 If you move by ~ 4, you will get a more accurate value. At this time, the detection speed of the camera 100 is 67ms, the average speed of the detection speed is 12 frames per second by using an input means that supports the speed of 15 frames per second, the motion detection of the moving object uses a line matching method.

According to the present invention, such a line matching method may be performed on one line or the difference value may be obtained using several lines, and the computer system 120 may be obtained by a line matching operation of Equation 2 below. Among the difference values stored in the memory of), the value which shows the smallest difference value is regarded as the movement of the corresponding line.

Sj = min ( Di ) In Equation 2 above, Sj represents a minimum value among the difference values of all pixels in the j-th line.

For example, when comparing the first line of the current video image with the first line of the previous video image for detecting the movement of each line using line matching of the real-time detection method of the moving object, the current video image is left and right While moving, the image difference of each movement is calculated and stored in the memory of the computer system 120. Among these, the movement when the difference is smallest is determined as the movement of the corresponding line, and may be performed at regular intervals of, for example, 5 lines without applying to all lines to improve performance.

On the other hand, in order to determine the motion of the entire video image, the most frequently detected motion among the motions calculated for each line of the video image is determined as the overall motion of the corresponding image. Equation 3 is an equation for determining the pan and tilt of the corresponding video image in two directions of the x-axis and y-axis of the video image in order to detect a moving object of the video image according to the present invention.

In Equation 3 above, P (Sy) represents values at which the difference values for the pixels of all lines are minimum, and Gshiftx is defined as a pan value, which is the most probable value among the above minimum values. Similarly, the tilt value can be obtained by changing Sx instead of Sy in P (Sy) in Equation 3 above and shifty instead of shiftx in Gshiftx.

As shown in FIG. 6 for detecting the motion of the entire video image of the real-time detection method of the moving object according to the present invention, the motion is obtained for each line, and the most frequently generated value (Pan 9, tilt -3) is obtained from the corresponding video image. Considering the overall motion of, we can eliminate the effects on moving objects in the video image.

When the coordinate calculation is completed in step S40, the pan / tilt of the pan / tilt driver 110 is controlled (step S50), and the motion of the detected pan / tilt is applied to the current video image to detect the moving object. The difference between the two images is obtained by using the distance that the camera moves from the previous frame to the current frame, so that a simple detection of the moving object can be performed. For example, if the pan movement is 5 to the right and the tilt movement is 4 to the bottom, the difference between the two images is obtained by adding the corresponding value to the position of each pixel of the previous image to compare it with the current frame In other words, the point where the difference in the video image value is large is indicated as motion. This reduces the amount of computation and time required to perform the conventional inverse transform, and thus can detect a moving object in real time as shown in FIG. 6 for detecting motion of the entire video image of the real-time detection method of the moving object according to the present invention. . When the pan / tilt is controlled in step S50, the camera 100 automatically tracks and monitors the image target for the next process (step S60).

As described above, in the detailed description of the present invention, specific embodiments have been described, but various modifications can be made without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined by the claims below and equivalents thereof.

According to the real-time detection method of the moving object according to the present invention the following advantages occur.

That is, camera pan / tilt recognition is searched for each line rather than motion search for the entire area, and the moving object is detected using this to reduce the amount of calculation and improve the moving detection speed, without using the video image itself. The boundary portion can be used to perform accurate detection.

Claims (7)

In a method of detecting a moving object in a camera moving with a pan / tilt driver, Obtaining a plurality of video images using the moving camera; Extracting edge components of each of the plurality of video images to detect the moving object within the plurality of video images; Comparing the edge components extracted from each of the plurality of video images with each other; Calculating coordinates of the moving object by comparing the edge components; And And adjusting the movement of the pan / tilt driver using the calculated coordinates. The method of claim 1, wherein the plurality of video images And at least two video images before and after a time point for detecting the movement of the camera. The method of claim 1, wherein the step of comparing the edge components with each other The camera moves each of the video images left and right, up and down, and stores the difference value of the edge component, and regards the smallest value among the difference values of the stored edge components as the movement of the corresponding edge component. Detection method. The method of claim 1, The detection speed of the camera is 67ms, characterized in that the moving object. The method of claim 1, The detection speed of the camera is a real-time detection method of a moving object, characterized in that the average speed of the detection speed is 12 frames per second by using an input means supporting a speed of 15 frames per second. The method of claim 1, The motion detection of the moving object is performed using a line matching method. delete