JP2003219422A - Method of detecting motion - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of detecting motion by which processing including noise removal can be performed with a small memory capacity and, at the same time, which can be loaded easily in equipment, because the geometric features of regions can be obtained by this simple method. <P>SOLUTION: This method of detecting motion includes a step of taking the differential amount in luminance component at every pixel between images and taking out the area of one value in a threshold processed image as a differential area, a step of adding a direction code indicting a direction to the borderline of the differential area, and a step of finding one or more values for the area, longitudinal and transversal lengths, or the peripheral length of the differential area. This method also includes a step of deciding that a motion is detected when the area, longitudinal and transversal lengths, or peripheral length of the differential area is equal to or larger than a prescribed size. <P>COPYRIGHT: (C)2003,JPO

Description

Description: BACKGROUND OF THE INVENTION [0001] 1. Field of the Invention [0002] The present invention relates to a motion detecting method for detecting a moving part from a moving image, and more particularly to an image input by a video camera, a camera connected to a PC, or the like. The present invention relates to a technique applied to a system for detecting, recording, transmitting, and recognizing a moving part from a moving object. [0002] Conventionally, many methods have been proposed as a motion detection method for detecting a moving part from a moving image. For example, a method of detecting a motion by performing region labeling by taking a difference between each image of a moving image and a difference from a background image, or dividing a moving image into a plurality of blocks and performing matching between the blocks. There was a way to detect movement. [0003] Conventional motion detection methods are often performed on a PC having sufficient memory capacity and processing capability, and thus consume a large amount of memory. There was no particular problem even if the processing amount was large.
On the other hand, as devices such as mobile terminals and mobile phones equipped with cameras have become widespread, the possibility that motion detection will be performed even with such mobile terminals and mobile phones has increased. However, the conventional motion detection method that consumes a large amount of memory or requires a large amount of processing as in the related art has been difficult to mount on devices such as mobile terminals and mobile phones. Also,
Since the memory consumption fluctuates depending on the distribution state of the motion area, processing with a limited memory amount may not be possible. There has been a demand for a method capable of detecting motion. For example, as a conventional motion detection method, there is a method of performing labeling. However, this method requires a memory area of the same size as an input image for labeling. Depending on the number of areas, the data storing the label may be 1 byte or more. In addition, a memory stack area for area search and label integration is required. Further, when there is noise in the image to be detected, it is necessary to eliminate the noise to increase the detection accuracy. Furthermore, there is a need for a motion detection method that can be applied to a wider range such that the position and shape characteristics of a moving area can be acquired in addition to whether or not there is a moving part in a moving image. SUMMARY OF THE INVENTION The object of the present invention is to obtain area information by extracting a contour line of a difference image for the purpose of enabling high-speed motion detection with a limited memory. Attention is paid to obtaining the amount and distribution of motion from the area information to perform motion detection. Since a pixel value changes in a moving part, the pixel value is extracted as a difference area, and then a contour line of each area is extracted. In the process of extracting a contour line, noise is removed based on shape characteristics such as the vertical and horizontal sizes, areas, and perimeters of the difference region, thereby improving detection accuracy. Further, a memory area required for performing contour line extraction is a memory area of data of the same size as the input image and 4 bits. This is constant regardless of the number of regions,
Since a stack memory area for label integration and the like required in the conventional labeling processing is not necessary, processing with a limited memory amount is possible. It is also possible to roughly detect motion by reducing the size of the original image. In this case, the amount of memory and the amount of processing can be further reduced. Further, since the shape characteristics of all the regions can be obtained only by scanning the difference region image once and extracting the contour lines, the processing can be performed with a small amount of calculation.
It is easy to implement because it is a simpler method than labeling, such as labeling required for labeling is not necessary. For this purpose, there is provided a motion detecting method for detecting a motion from a plurality of images constituting a moving image based on a difference amount of a luminance component between the images, and calculating a difference amount of a luminance component for each pixel between the images. Extracting a region of one value in an image obtained by binarizing the difference amount by a predetermined threshold value as a difference region; adding a direction code indicating a direction to a contour line of the difference region; Tracking the contour along a code to determine any one or more values of the area of the difference region, the length and width of the difference region, and the length of the periphery of the difference region; Determining that a motion has been detected when any of the area of the difference area, the vertical and horizontal lengths of the difference area, and the perimeter of the difference area is equal to or greater than a predetermined size. The law provides. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The specific contents of a motion detection method according to the present invention will be described below in detail with reference to the drawings.
FIG. 1 is a configuration diagram of a motion detection system to which a motion detection method according to the present invention is applied. The system includes an image input device 1 for inputting moving images, such as a video camera and a PC camera, a storage device 2 for storing image data and programs,
A motion detection device 3 that performs motion detection from an image input from the image input device 1, an input device 4 such as a mouse or a keyboard,
It comprises an image display device 5 for displaying moving images and the like, and an output device 6 such as a printer for printing moving images and the like. Next, the procedure of the motion detection method according to the present invention will be described with reference to the flowchart shown in FIG. The image data input by the image input device 1 is sequentially stored in the storage device 2. FIG. 3 shows an example of the input image at this time. The size of the input image and the number of frames per second depend on the performance of the camera and the performance of the image input unit.
A description will be given using an example of a 240 × 176 image input at 00 ms intervals. Since the brightness (brightness) of the moving part of the image changes, first, the absolute value of the difference between the pixel values of the images is calculated (step S1). If the original pixel value is in the range of 0 to 255, the difference value is also 0 to 255.
At this time, since there is a small change in the pixel value other than the movement due to the characteristic of the image input device 1 or a change in illumination, a pixel having a difference value equal to or more than a certain threshold value is regarded as a portion where the movement has occurred. That is, the moving part and the non-moving part are binarized based on the threshold value (step S2). The threshold value is set to an appropriate value depending on the performance of the image input apparatus 1 and the environment of the object to be photographed (the degree of fluctuation in lighting and brightness). FIG. 4A shows a frame image 1, FIG.
4B is a frame image 2, and FIG. 4C is a binarized image obtained by taking the difference between the frame image 1 and the frame image 2 and setting the threshold value to 16. In FIG. 4C,
The black pixel portion is a pixel whose difference value is equal to or greater than the threshold value, and this black pixel portion is called a difference region. As described above, it is determined whether or not there is movement between two images based on the area of the difference area. To calculate the total area, the number of black pixels in the figure, which is a difference area, may be counted. However, in this method, even a small area such as a fine noise component is included, so that FIG.
5C and a frame image 2 shown in FIG. 5B when there is actually no motion and a noisy image appears when the difference is calculated, as shown in FIG. May be erroneously detected as having motion. Therefore, the size and area in the vertical and horizontal directions are obtained for each difference area by the following method, and only the area having a certain size or more is summed up as a motion area to obtain a difference amount. In addition, in addition to the vertical and horizontal size and area, shape characteristics such as a perimeter and a region center of gravity can be obtained as shape characteristics. In order to extract the outline of the area, first, a data area having the same size as the image is prepared, and a direction code is added so as to track around the black pixel. Since the direction code is ON / OFF in four directions, up, down, left, and right, it is 4-bit data (FIG. 6). Next, the direction code is searched from the upper left of the image to find the starting point, and the contour is traced from the starting point along the direction code. When returning to the starting point, one area is set (step S3). In the course of this tracking, the size, area (number of pixels), and perimeter of this area are determined. It is also possible to determine the center of gravity of the region from the vertical and horizontal black pixel positions. (1) The vertical and horizontal sizes are obtained from the maximum and minimum coordinate values of the tracking point (FIG. 7A). (2) The area (the number of pixels) can be obtained by setting the initial value to 0, adding the x coordinate value when tracking upward, and subtracting the x coordinate value when tracking downward (FIG. 7B). ). (3) The perimeter is the number of tracking points (FIG. 7 (c)). The amount of motion of the entire image is the sum of the areas (number of pixels) of each region. However, if the vertical and horizontal sizes are smaller than the specified size, they are not added to the total as noise. If there is a hole in the difference area as shown in FIG. 8, the following processing is performed. Since the tracking direction of the hole in the area is opposite to that of the outside of the area, the area (number of pixels) of the area becomes a negative value in that case. When the above-mentioned negative value is added to the area of the entire image, it is subtracted from the area.
In this case, if the vertical and horizontal sizes are smaller than the specified size, they are excluded from the calculation. According to this method, even in the case of a difference area having a hole, the area and the motion amount of the area can be accurately obtained.
Further, fine noise components in the motion area can be removed (step S4). FIG. 9 shows the result of removing noise by this method. The direction code is turned off for the tracking point at which the contour has been tracked. By sequentially searching for the starting point and scanning to the lower right, all regions can be extracted and the shape characteristics can be obtained. When the extraction of all the regions is completed, the total area of the moving regions is obtained, so that it is possible to determine whether or not there is any movement based on this amount. By adding a process such as recording an image when there is a motion according to this data, application to a security image monitor becomes possible. Further, by separately recording and processing the position, size and shape characteristics of each area, it is possible to apply to applications such as tracking of moving objects and image recognition. As described above, according to the motion detection method of the present invention, it is possible to perform processing including noise elimination with a small amount of memory. In addition, by performing the region extraction processing by contour tracing, a region shape feature can be obtained by a simpler method as compared with conventional labeling, so that mounting on a device is facilitated. Further, by using the contour line, it is possible to reduce the amount of processing required for detection, and it is possible to mount the apparatus on a device having a small calculation capability such as a portable terminal or a portable telephone.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a configuration of a motion detection system to which a motion detection method according to the present invention has been applied. FIG. 2 is a flowchart showing one embodiment of a motion detection method according to the present invention. FIG. 3 is a diagram showing an example of an input image for detecting a motion by the motion detection method according to the present invention. FIG. 4 is a diagram showing an example of an image in which a motion is extracted by a difference value applied to the motion detection method according to the present invention. FIG. 5 is a diagram showing a state where there is no motion in an image from which motion has been extracted based on a difference value applied to the motion detection method according to the present invention. FIG. 6 is a diagram showing a contour line extraction method in the motion detection method according to the present invention. FIG. 7 is a diagram showing a shape characteristic of a region from which a contour is extracted in the motion detection method according to the present invention. FIG. 8 is a diagram showing a case where there is a hole in an area where a contour line is extracted in the motion detection method according to the present invention. FIG. 9 is a diagram showing a state of noise removal in the motion detection method according to the present invention. [Description of Signs] 1 Image input device 2 Storage device 3 Motion detection device 4 Input device 5 Image display device 6 Output device

Claims (1)

1. A motion detecting method for detecting a motion from a plurality of images constituting a moving image based on a difference amount of a luminance component between the images, the method comprising: detecting a luminance component for each pixel between the images. Take the difference amount of
Extracting a region of one value in the image obtained by binarizing the difference amount by a predetermined threshold value as a difference region; adding a direction code indicating a direction to an outline of the difference region; By tracking the contour along
Determining any one or more values of the area of the difference area, the length and width of the difference area, and the length of the periphery of the difference area; and the area of the difference area and the length and width of the difference area. Determining that a motion has been detected when any of the lengths of the periphery of the difference area is equal to or greater than a predetermined size.