JP4470342B2 - Image processing method and image processing apparatus using the same - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing method for automatically detecting the movement of an object such as a person or a vehicle in a predetermined monitoring area, and an image processing apparatus using the image processing method.
[0002]
[Prior art]
As this kind of image processing method, a background difference method and an inter-frame difference method have been conventionally provided.
[0003]
First, the background difference method will be described with reference to FIGS. In this method, an image of the monitoring area A (see FIG. 11A) when no moving object exists in advance is captured and stored as a background image. Thereafter, the image of the monitoring area A is sequentially picked up, the difference from the background image is obtained, and it is determined that the area where the difference has occurred is the part where the object not in the background has entered. For example, when a person H enters the monitoring area A, if the difference between the image of the monitoring area A at this time (see FIG. 11B) and the background image is taken, the binary difference shown in FIG. Since an image is obtained, it can be determined from this image that the person H has entered.
[0004]
Next, the interframe difference method will be described with reference to FIGS. In this method, images of the monitoring area A are sequentially captured, and a moving object is detected by taking a difference between two images captured at different times. FIG. 12A shows an image of the monitoring area A taken at a certain time, and FIG. 12B shows an image of the monitoring area A taken at a certain time thereafter. Since the difference image becomes a difference image and a binary difference image as shown in FIG. 12C is obtained, it can be detected from this image that some object is moving.
[0005]
[Problems to be solved by the invention]
Among the above-described image processing methods, the former method (background difference method) needs to capture an image (background image) of a monitoring area in which no person exists in advance, but there is a person in the monitoring area. The operator must visually confirm that the image has not been captured, and the background image cannot be captured automatically. In addition, the operation of capturing the background image has to be performed at the actual installation site, and there is a problem that it takes time and effort for the imaging operation.
[0006]
In order to automatically capture the background image, it is conceivable that the monitoring area is imaged for a long time and the average value is used as a background image. There is a problem that people are buried in the background image. In addition, it is possible to take a method of imaging the monitoring area for a long time and sequentially updating the background image for each area, with the background of the area with little or no change from the captured image. When a person is stationary all the time, there is a problem that the person is buried in the background image as described above.
[0007]
On the other hand, in the latter method (inter-frame difference method), the difference between two images taken at different times is obtained, and the moving object is extracted with double images of the object. Therefore, there is a problem that the means for determining whether or not the extracted object is a person is complicated.
[0008]
In addition, in this method, the changed part is extracted from the two images, so when only a part of the human body is moving, only this part can be extracted, and the shape of the human body itself is extracted. I couldn't. For example, when a person is sitting or sleeping in the surveillance area, if only the limbs move and the torso is stationary, only the moving limbs are extracted, and the hand and feet are separated. Therefore, it is necessary to determine whether the extracted hand and foot are those of the same person or different persons, and this procedure has a problem that it is considerably complicated.
[0009]
The present invention has been made in view of the above problems, and an object of the present invention is to provide an image processing method and an image processing method that can detect the entire search target without having to capture a background image in advance. It is to provide an image processing apparatus used.
[0010]
[Means for Solving the Problems]
  In order to achieve the above object, according to the first aspect of the present invention, there is provided an image processing method for detecting the presence or absence of a search object such as a person or a vehicle in a predetermined monitoring area and its movement, wherein the monitoring area has a predetermined timing. Grayscale imageTo the cameraImageAnd input from the cameraCreate a contour image showing the contour of an object from the grayscale imageprocessingAnd creating a contour image showing the contour of the object from the grayscale imageprocessingAnd an image of a stationary object that does not move between the first contour image created last time and the second contour image created this timeprocessingAnd the image of the moving part of the search object moved between the second contour image and the still object image extracted this timeprocessingWhen,Processing to extract the part that exists only in the static object image extracted this time but does not exist in the previously extracted static object image as the stop part image of the search objectAnd an image of the search area including the search object from the extracted moving part and stopping part imagesprocessingAnd the search area image and this timeSecondPerforms matching processing between the contour image and the template image to determine the existence of the search object and its movementprocessingAndLet the CPU do itIn the search object, the image of the moving part and the image of the stopped part are obtained, and the whole image of the search object is obtained from both images. As compared with the case where only the moving part is extracted, the search object can be easily discriminated, and the search object image is extracted twice as in the inter-frame difference method. Therefore, processing such as template matching can be easily performed. Furthermore, since it is not necessary to capture a background image in a state where no search object exists unlike the background difference method, it is possible to eliminate the trouble of capturing the background image.
[0011]
  Moreover,As the image of the stop part of the search object, the part that exists only in the image of the stationary object extracted this time but not in the previously extracted stationary object image is extracted.BecauseThe image of the stop part of the search object can be extracted by a simple method.
[0012]
  Claim2In the invention of claim1In this invention, an area including the contour of the object is set as a search area for performing template matching in the image obtained by calculating the logical sum of the moving part image and the stop part image extracted this time.Have the CPU do the processingAnd claims1In addition to the operation of the present invention, since the logical sum of the moving part image and the stop part image is obtained, the entire shape of the search object can be determined, and the correlation value with the template image is obtained accurately. be able to.
[0013]
  Claim3In the invention of claim1In this invention, the image obtained by calculating the logical sum of the moving part image and the stop part image extracted this time is used as a first search image for template matching, and the contour of the object in the first search image is defined as the first search image. While calculating the correlation value with the template image for the included region, the contour image created this time is used as the second search image for template matching, and the vicinity of the previous detection position of the search object in the second search image is used as the search region. The correlation value with the template image is obtained in the search area, the area having the highest correlation value and higher than a preset threshold is set as the detection position of the search object, and the next template image is set based on the image at the detection position. DoHave the CPU do the processingAnd claims1In addition to the operation of the present invention, the image obtained by calculating the logical sum of the image of the moving part extracted this time and the image of the stop part is used as the first search image, and the contour image created this time is used as the second search image. Since template matching is performed, the presence or absence of the search object can be reliably detected regardless of whether the search object is moving or stopped.
[0014]
  Claim4In the invention of claim1In the present invention, among the regions existing in the previously extracted stop portion image, the regions not present in the currently extracted stop portion and moving portion images are added to the currently extracted stop portion image.Have the CPU do the processingAnd claims1In addition to the action of the invention of the present invention, when the stop part that was stopped last time continues to stop, this stop part does not appear in either the stop part extracted this time or the image of the moving part. The shape of the search object can be accurately detected by adding the region not present in the image of the stop part and the moving part extracted this time to the image of the stop part extracted this time.
[0015]
  Claim5In the present invention, claims 1 to4In the invention, a portion where no difference exists between the first contour image and the second contour image is extracted as a still object image.Have the CPU do the processingClaims 1 to4In addition to the operation of the present invention, an image of a stationary object can be extracted by a simple process.
[0016]
  Claim6In the invention of claim 1, in the inventions of claims 1 to 5, a part where a difference exists between the second contour image and the image of the stationary object extracted this time is extracted as the image of the moving part of the search object.Have the CPU do the processingIn addition to the effects of the first to fifth aspects of the invention, the moving part image can be extracted by a simple process.
[0017]
  Claim7In the invention, in the image processing apparatus for detecting the presence or absence of a search object such as a person or a vehicle in a predetermined monitoring area and its movement, a means for capturing a grayscale image of the monitoring area at a predetermined timing, and the above grayscale image Means for creating a contour image showing the contour of the object, means for extracting an image of a stationary object that has not moved between the first contour image created last time and the second contour image created this time, Means for extracting, from the contour image of 2 and the image of the stationary object extracted this time, an image of the moving part of the search object that has moved between the two,Means for extracting, as an image of the stop part of the search object, a portion that is not present in the previously extracted stationary object image but is present only in the stationary object image extracted this time;Means for creating an image of the search area including the search object from the image of the moving part and the stop part extracted this time, and the presence / absence of the search object by performing a matching process between the image of the search area and the contour image created this time and the template image And a means for judging the movement of the search object, and obtaining an image of the moving part and an image of the stopped part of the search object, respectively. Compared to extracting only moving parts as in the case of using the inter-frame difference method, the search object can be easily discriminated, and the inter-frame difference method is used. As described above, since the image of the search object is not extracted twice, it is possible to provide an image processing apparatus that can easily perform processing such as template matching. Furthermore, since it is not necessary to previously capture a background image in a state where no search object exists as in the case of using the background difference method, it is possible to provide an image processing apparatus that eliminates the trouble of capturing a background image.Moreover, as the image of the stop part of the search object, the part of the search object that is not present in the previously extracted still object image but is present only in the still object image extracted this time is extracted. Can be extracted in a simple way.
[0018]
DETAILED DESCRIPTION OF THE INVENTION
A first embodiment of the present invention will be described with reference to FIGS. This image processing apparatus automatically detects the presence / absence and movement of a person H (search object) in a predetermined monitoring area A (for example, a room in a building). FIG. 1 is a schematic configuration diagram of this image processing apparatus. For example, a camera 1 such as a CCD camera that is installed on a ceiling of a room and picks up a black and white grayscale image in the room, and an analog video input from the camera 1. Supplied from an A / D converter 2 that converts a signal into a digital signal, a CPU 3 that detects the presence / absence and movement of a person H by processing the output signal of the A / D converter 2, and a DC power supply (not shown) And a power supply circuit 4 for supplying power to the camera 1 by converting the power supply voltage to a DC voltage having a desired voltage value.
[0019]
An image processing method of the image processing apparatus will be described with reference to FIGS. 3 (a) is an input image at time T0 when the person H is falling down, FIG. 3 (b) is an input image at the time T1 at the moment when the person H has fallen (lie on the floor), and FIGS. d) respectively shows the input images at times T2 and T3 in a state where the person H is standing still with the person H lying down, and the person H is stationary between the times T1 to T3. 4A to 4D show contour images at times T0 to T3, FIGS. 5A to 5D show still object extraction images at times T0 to T3, and FIGS. d) is an extracted image of the moving part of the search object at times T0 to T3, FIGS. 7A to 7D are extracted images of the stop part of the search object at times T0 to T3, and FIGS. d) shows search images at times T0 to T3, and FIGS. 8E, 8G, 8I, and 8K show template images t01, t11, t21, and t31 used when template matching is performed at each time. (F), (h), (j), and (l) respectively show template images t02, t12, t22, and t32 that reflect the results of template matching. 4 to 8 are schematic diagrams, and the lines indicating the outline of the object are illustrated with the same thickness lines, but in reality, there are thin portions and thick portions, so that they disappear in the process. Part and remaining part occur.
[0020]
The camera 1 captures black and white grayscale images in the monitoring area A at predetermined time intervals, and the video signal of the camera 1 is converted into a digital signal by the A / D converter 2 and sequentially input to the CPU 3 (FIG. 2). Step S1). When a black and white grayscale image from the camera 1 is input to the CPU 3, the CPU 3 creates a differential image from the black and white grayscale image and binarizes it to detect the boundary line (edge) of the object. That is, in the part corresponding to the object in the image, its characteristics, for example, the density is uniform, and it is considered that the density changes suddenly between different objects or parts of the object, so search for a place where the density changes rapidly. Thus, the boundary line (edge) of the object can be detected. Also, if two different objects overlap, the density becomes discontinuous at the boundary between them, and the amount of change (gradient) in density is large, so by detecting the location where the density gradient is large, The boundary line of the object can be detected. In general, it is only necessary to perform spatial differentiation in order to detect discontinuity of light and dark, and by obtaining a difference between adjacent pixels in a black and white light image captured at a certain time, a differential value of density, that is, a gradient (gradient) is obtained. Can be detected. Note that various methods are provided as edge detection methods, and a method using a Sobel arithmetic expression is generally performed.
[0021]
That is, the CPU 3 performs image processing on the image data of the black and white image and obtains the gradient value of the density. If the gradient value is higher than the threshold value, the gradient value is compared with the predetermined threshold value. If the contour portion is equal to or less than the threshold value, it is determined as a flat portion (binarization process). For example, FIGS. 3A to 3D are obtained from black and white grayscale images (input images) input from the camera 1 at times T0 to T3, and FIGS. 4A to 4D are obtained from black and white grayscale images at each time. Each contour image is shown, and the solid line in FIGS. 4A to 4D shows the contour of the object. In this embodiment, since it is difficult to distinguish between the body part of the human body H and the face and the background, the contour image of FIGS. 4A to 4D cannot detect the body part and the face outline. The image looks like the head and limbs are separated.
[0022]
Next, the CPU 3 obtains the logical product of the contour image (first differential image) created last time and the contour image (second differential image) created this time, thereby obtaining the first differential image and the second differential image. A portion where there is no difference between them (that is, a portion that has not changed) is extracted, and an image of a stationary object (stationary object) from the previous imaging time point to the current imaging time point (FIG. 5A ) To (d)) (step S2 in FIG. 2). For example, at time T2, the CPU 3 obtains a logical product of the contour image at time T1 (see FIG. 4B) and the contour image at time T2 (see FIG. 4C), so that the stationary image does not move during this time. An object image (see FIG. 5C) is extracted.
[0023]
Further, the CPU 3 obtains a difference between the contour image (second differential image) created this time and the image of the stationary object extracted in step S2, and thus a portion where the difference exists between the two images (that is, moved). (Part)) is extracted, and an image (see FIGS. 6A to 6D) of the moving part of the search object that has moved between the previous imaging time point and the current imaging time point is created (step S3 in FIG. 2). . For example, at time T1, the CPU 3 obtains the difference between the contour image at this time (see FIG. 4B) and the image of the stationary object (see FIG. 5B), thereby moving the moving part image (FIG. b) is extracted. At time T1, since the person H's head and part of the hand and the left foot are stationary, the image is such that the part where the head and hand are moving and the part of the right foot remain.
[0024]
Further, the CPU 3 obtains a difference between the image of the stationary object extracted last time and the image of the stationary object extracted this time, so that it does not exist in the image of the stationary object extracted last time, but only the image of the stationary object extracted this time. An existing portion is extracted, and an image of only the stop portion of the search object (see FIGS. 7A to 7D) is extracted from the image of the still object extracted this time (step S4 in FIG. 2). For example, at time T2, the CPU 3 obtains a difference between the still object image extracted at time T2 (see FIG. 5C) and the still object image extracted at time T1 (see FIG. 5B). The image of only the stop part of the search object (see FIG. 7C) is extracted.
[0025]
When extracting the stop part of the search object, the CPU 3 extracts the stop part extracted this time and the stop part extracted this time and the area not present in the image of the movement part among the areas existing in the previously extracted stop part image. You may make it add to the image of a site | part. For example, at time T3, the stop part stopped at time T2 continues to stop, so this stop part does not appear in any of the stop part and moving part images extracted this time, and as a result a search image (FIG. 8 ( In (d) see), nothing appears, but out of the areas existing in the previously extracted stop part image, the stop part extracted this time and the non-existing area in the moving part image are displayed in the stop part image extracted this time. If added, the shape of the search object (person H) can be accurately detected.
[0026]
Thereafter, the CPU 3 calculates the logical sum of the image of the moving part extracted in step S3 and the image of the stop part extracted in step S4, and further removes the noise component, thereby performing template matching to perform a search object (human body). Search images (see FIGS. 8A to 8D) are created (step S5 in FIG. 2).
[0027]
Here, a method for detecting presence / absence of a search object and its position by performing template matching will be described below.
[0028]
First, the CPU 3 extracts a search area where a search object is considered to exist from the search image created in step S5. FIG. 9A shows an example of the search image created in step S5. As shown in FIG. 9, the boundary line E of the search object (person H) is divided into small pieces in the search image. There is a problem that it is not known in which part the person H who is the search object exists. Therefore, the CPU 3 performs the expansion process on the search image obtained in step S5 to expand the boundary line E divided into small pieces, thereby forming a block B as shown in FIG. 9B. A circumscribed rectangle circumscribing the block B in this image is obtained. When the circumscribed rectangles including the blocks B are close to each other to some extent, these circumscribed rectangles are combined into one, and search areas C1 to C3 are determined from the circumscribed rectangles thus obtained. Yes. When determining the search areas C1 to C3, an area larger than the circumscribed rectangle circumscribing the block B is set in order to allow a certain margin. For example, in the search image shown in FIG. 9 (a), only a part of the boundary line of the human body appears and the boundary line E of the head and legs is shredded. The boundary line E thus formed can be integrated to restore the boundary line of the human body. In addition, when a plurality of blocks B are formed by expanding the cut border line E of the human body, since the circumscribed rectangles including these blocks B are close to each other, the search region C1 including the human body is the final Thus, they are grouped into one area.
[0029]
The CPU 3 sets a search area in the search image by the method as described above, and obtains a correlation value between the two by performing a matching process between the image in the search area and a predetermined template image (see FIG. 2 step S6). For example, at time T1, the CPU 3 superimposes the template image t11 shown in FIG. 8G on the search area of the search image (see FIG. 8B) at this time, and shifts the position of the template image t11 pixel by pixel. The correlation value is detected. If the highest correlation value in the search area exceeds a preset threshold, it is determined that a person exists, and the search area corresponding to the correlation value is set at a position where the person exists. ing.
[0030]
Further, the CPU 3 performs template matching for the contour image created at the same time as template matching for the search image. In this case, the CPU 3 sets the vicinity of the previous human body detection position in the contour image as a search region (not shown), performs a matching process between the image in the search region and the template image, and obtains a correlation value between them. Yes. For example, at the time T1, the CPU 3 sets the vicinity of the previous human body detection position in the search region for the contour image at this time (see FIG. 4B), and then superimposes the template image t11 on the image in the search region, The correlation value is detected while shifting the position of t11 by one image. If the highest correlation value in the search area exceeds a preset threshold, it is determined that a person exists, and the search area corresponding to the correlation value is set at a position where the person exists. ing.
[0031]
Here, when the correlation value obtained by performing the template matching for the search image is higher than the correlation value obtained by performing the template matching for the contour image, the CPU 3 determines that the human body is moving, and there is a person. The image of the search area in the determined search screen is set as the next template image, and the detection position of the person H is updated (step S7 in FIG. 2). For example, at time T1, the correlation value obtained by performing the template matching on the search image becomes higher than the correlation value obtained by performing the template matching on the contour image, so the image of the search area set in the search image is used as the template image. Set to t12.
[0032]
On the other hand, if the correlation value obtained by performing the template matching on the contour image is higher than the correlation value obtained by performing the template matching on the search image, the CPU 3 determines that the human body is stopped, and the template image is updated. Not performed. For example, at time T3, since the person H remains stationary from time T1 to T3, nothing appears in the search image. As a result, the correlation value obtained by performing the template matching on the contour image becomes the template matching on the search image. Therefore, the template image is not updated and the current template image t31 is used as the next template image t32. FIGS. 10A to 10C show template images ta to tc used when the power is turned on, and template images having a shape that symbolizes the head and torso of the human body H are used.
[0033]
As described above, in the image processing apparatus according to the present embodiment, an image of a moving part and an image of a stopped part of the search object (person H) are obtained, and the entire search object is obtained from both images. Compared to the case of extracting only moving parts as in the inter-frame difference method, the search object can be easily identified, and the search is performed as in the inter-frame difference method. Since object images are not extracted twice, processing such as template matching can be easily performed. Furthermore, since it is not necessary to capture a background image in a state where no search object exists unlike the background difference method, it is possible to eliminate the trouble of capturing the background image.
[0034]
【The invention's effect】
  As described above, the invention of claim 1 is an image processing method for detecting the presence or absence of a search object such as a person or a vehicle in a predetermined monitoring area and its movement, and a grayscale image of the monitoring area at a predetermined timing. TheTo the cameraImageAnd input from the cameraCreate a contour image showing the contour of an object from the grayscale imageprocessingAnd an image of a stationary object that does not move between the first contour image created last time and the second contour image created this timeprocessingAnd the image of the moving part of the search object moved between the second contour image and the still object image extracted this timeprocessingWhen,Processing to extract the part that exists only in the static object image extracted this time but does not exist in the previously extracted static object image as the stop part image of the search objectAnd an image of the search area including the search object from the extracted moving part and stopping part imagesprocessingAnd the search area image and this timeSecondPerforms matching processing between the contour image and the template image to determine the existence of the search object and its movementprocessingAndLet the CPU do itIn the search object, the image of the moving part and the image of the stopped part are obtained, and the whole image of the search object is obtained from both images. As compared with the case where only the moving part is extracted, the search object can be easily discriminated, and the search object image is extracted twice as in the inter-frame difference method. Therefore, there is an effect that processing such as template matching can be easily performed. In addition, unlike the background subtraction method, there is no need to capture a background image in a state where no search object exists in advance, so that it is possible to eliminate the trouble of capturing the background image.
[0035]
  Moreover,As the image of the stop part of the search object, the part that exists only in the image of the stationary object extracted this time but not in the previously extracted stationary object image is extracted.BecauseThere is an effect that an image of a stop portion of the search object can be extracted by a simple method.
[0036]
  Claim2The invention of claim1In this invention, an area including the contour of the object is set as a search area for performing template matching in the image obtained by calculating the logical sum of the moving part image and the stop part image extracted this time.Have the CPU do the processingAnd claims1In addition to the operation of the present invention, since the logical sum of the moving part image and the stop part image is obtained, the entire shape of the search object can be determined, and the correlation value with the template image is obtained accurately. There is an effect that can be.
[0037]
  Claim3The invention of claim1In this invention, the image obtained by calculating the logical sum of the moving part image and the stop part image extracted this time is used as a first search image for template matching, and the contour of the object in the first search image is defined as the first search image. While calculating the correlation value with the template image for the included region, the contour image created this time is used as the second search image for template matching, and the vicinity of the previous detection position of the search object in the second search image is used as the search region. The correlation value with the template image is obtained in the search area, the area having the highest correlation value and higher than a preset threshold is set as the detection position of the search object, and the next template image is set based on the image at the detection position. DoHave the CPU do the processingAnd claims1In addition to the operation of the present invention, the image obtained by calculating the logical sum of the image of the moving part extracted this time and the image of the stop part is used as the first search image, and the contour image created this time is used as the second search image. Since template matching is performed, there is an effect that the presence or absence of the search object can be reliably detected regardless of whether the search object is moving or stopped.
[0038]
  Claim4The invention of claim1In the present invention, among the regions existing in the previously extracted stop portion image, the regions not present in the currently extracted stop portion and moving portion images are added to the currently extracted stop portion image.Have the CPU do the processingAnd claims1In addition to the action of the invention of the present invention, if the stop part that was stopped last time continues to stop, this stop part does not appear in either the stop part extracted this time or the image of the moving part. Among the regions existing in this image, there is an effect that the shape of the search object can be accurately detected by adding the region not present in the image of the stop region and the moving region extracted this time to the image of the stop region extracted this time.
[0039]
  Claim5The invention of claim 1 to claim 14In the invention, a portion where no difference exists between the first contour image and the second contour image is extracted as a still object image.Have the CPU do the processingClaims 1 to4In addition to the operation of the invention, there is an effect that an image of a stationary object can be extracted by a simple process.
[0040]
  Claim6According to the invention of any one of claims 1 to 5, as the image of the moving part of the search object, a portion where a difference exists between the second contour image and the image of the stationary object extracted this time is extracted.Have the CPU do the processingIn addition to the effects of the first to fifth aspects of the invention, there is an effect that the image of the moving part can be extracted by a simple process.
[0041]
  Claim7In the image processing apparatus for detecting the presence / absence of a search object such as a person or a vehicle in a predetermined monitoring area and its movement, means for capturing a grayscale image of the monitoring area at a predetermined timing, and the grayscale image Means for creating a contour image showing the contour of the object; means for extracting an image of a stationary object that has not moved between the first contour image created last time and the second contour image created this time; Means for extracting, from the contour image of 2 and the image of the stationary object extracted this time, an image of the moving part of the search object that has moved between the two,Means for extracting, as an image of the stop part of the search object, a portion that is not present in the previously extracted stationary object image but is present only in the stationary object image extracted this time;Means for creating an image of the search area including the search object from the image of the moving part and the stop part extracted this time, and the presence / absence of the search object by performing a matching process between the image of the search area and the contour image created this time and the template image And a means for judging the movement of the search object, and obtaining an image of the moving part and an image of the stopped part of the search object, respectively. Compared to extracting only moving parts as in the case of using the inter-frame difference method, the search object can be easily discriminated, and the inter-frame difference method is used. As described above, since the image of the search object is not extracted twice, it is possible to provide an image processing apparatus that can easily perform processing such as template matching. That. Furthermore, unlike the case where the background subtraction method is used, it is not necessary to capture a background image in a state where no search object exists in advance, so that it is possible to provide an image processing apparatus that eliminates the trouble of capturing the background image. .Moreover, as the image of the stop part of the search object, the part of the search object that is not present in the previously extracted still object image but is present only in the still object image extracted this time is extracted. Can be extracted in a simple way.
[Brief description of the drawings]
FIG. 1 is a schematic configuration diagram of an image processing apparatus according to an embodiment.
FIG. 2 is a flowchart illustrating the operation of the above.
FIGS. 3A to 3D are explanatory diagrams of an input image at each time.
4A to 4D are explanatory diagrams of a contour image at each time. FIG.
FIGS. 5A to 5D are explanatory diagrams of an extracted image of a stationary object at each time.
FIGS. 6A to 6D are explanatory diagrams of extracted images of moving objects at each time.
7A to 7D are explanatory diagrams of extracted images of a stopped object at each time.
FIGS. 8A to 8D are explanatory diagrams of a search image at each time, and FIGS. 8E to l are explanatory diagrams of a template image at each time.
FIGS. 9A and 9B are explanatory diagrams illustrating a procedure for creating a search area from a search image. FIGS.
FIGS. 10A to 10C are explanatory diagrams of template images when power is turned on.
FIGS. 11A to 11C are explanatory diagrams of an image processing method using a background difference method.
FIGS. 12A to 12C are explanatory diagrams of an image processing method using an inter-frame difference method.
[Explanation of symbols]
1 Camera
3 CPU
A monitoring area

Claims (7)

An image processing method for detecting the presence or absence of a search object such as a person or vehicle in a predetermined monitoring area and its movement,
A process of creating a contour image showing the outline of an object from the grayscale image inputted from a predetermined Rutotomoni camera is captured camera grayscale image of the monitor area in the timing, created this time and the first contour image previously created was from a second contour image, a process of extracting an image of a stationary object is not moving during this period, the and a second outline image and the image of the current extracted still object, moving part of the search object has moved in the meantime The process of extracting the image of the search object, the process of extracting the part that exists only in the image of the still object extracted this time but not in the image of the still object extracted last time, and the process of extracting this time Matching from the mobile site and stop site of the image processing for creating an image in the search region including the search object, the image and the second outline image and the template image created this search area Image processing method characterized by causing the process of determining presence or the movement of the search object by performing a process to the CPU. Let the CPU perform the process of setting the region including the contour of the object as the search region for template matching in the image obtained by calculating the logical sum of the image of the moving part and the image of the stopping part extracted this time. The image processing method according to claim 1, wherein: The image obtained by calculating the logical sum of the image of the moving part and the image of the stop part extracted this time is used as a first search image for template matching, and a template for the region including the contour of the object in the first search image is used. While calculating the correlation value with the image, the contour image created this time is used as the second search image for template matching, and the vicinity of the previous detection position of the search object in the second search image is set as the search region. A process of obtaining a correlation value with the template image, setting a region having the highest correlation value and higher than a preset threshold as a detection position of the search object, and setting a next template image based on the image of the detection position, The image processing method according to claim 1 , wherein the image processing is performed by a CPU . A feature of causing the CPU to perform processing for adding a region not present in the image of the stop part extracted this time and a part of the movement part extracted in the image of the stop part extracted this time to the image of the stop part extracted this time. The image processing method according to claim 1 . An image of a stationary object, the process of extracting the partial absence of the difference between the first outline image and the second outline image, either of claims 1 to 4, characterized in that causing the CPU 1 The image processing method according to item . An image of the moving part of the search object, the processing for extracting a portion difference was present between the second contour image and the image of the current extracted stationary object, 1 to claim, characterized in that to perform the CPU the image processing method according to any one of 5. In an image processing apparatus for detecting the presence or absence of a search object such as a person or a vehicle in a predetermined monitoring area and its movement, a means for capturing a gray image of the monitoring area at a predetermined timing, and an outline of the object from the gray image Means for creating a contour image to be shown, means for extracting an image of a stationary object that has not moved between the first contour image created last time and the second contour image created this time, and the second contour image Means for extracting the image of the moving part of the search object moved between this time and the image of the stationary object extracted this time, and only the image of the stationary object extracted this time that does not exist in the previously extracted stationary object image Means for extracting an existing portion as an image of a stop part of a search object, means for creating an image of a search area including the search object from the image of the moving part and stop part extracted this time, and search The image processing apparatus characterized by comprising a means for determining the image and presence and the movement of the search object by performing a matching process between the contour image and the template image created this range.

Images