JP4811289B2 - Image processing device - Google Patents

Description

  The present invention relates to an image processing apparatus.

  2. Description of the Related Art Conventionally, an image processing apparatus that detects a moving object based on a captured image is used for detecting a car as a moving object in a ticket machine in an unmanned parking lot, for example (see, for example, Patent Document 1).

  This type of image processing apparatus includes an imaging unit that periodically captures an image of a predetermined monitoring range, and a reference image as illustrated in FIG. 5A that is an image captured by the imaging unit and has no moving object. A storage unit that is stored, and a detection unit that detects the moving object M by comparing a latest image captured by the imaging unit, for example, as illustrated in FIG. 5B with a reference image. That is, the detection unit detects a difference from the reference image in the latest image as a moving object M as shown in FIG. 5C. For example, in a monitoring application, a part of the moving object is extracted to identify the moving object. For use in crime prevention, an alarm is issued when a moving object is detected.

  Here, the reference image is an image captured by the imaging unit when there is no moving object in the monitoring range due to the displacement of a stopped object such as a door or chair, or the appearance or disappearance of a stopped object such as a luggage or footprint. If they are different, there is a possibility of erroneous determination in which it is determined that there is a moving object even though there is no moving object.

Therefore, the image captured by the imaging unit is stored in the storage unit as the previous image until the next imaging, and even if the latest image and the reference image are different from each other, It has been proposed that it is determined that there is a displacement or appearance / disappearance of a stationary object, and the reference image is updated to the previous image. By adopting this configuration, it is possible to appropriately update the reference image in accordance with the displacement, appearance / disappearance of the stopped object, and to reduce the erroneous determination.
JP 2002-150272 A

  However, since the part where there was no difference between the latest image and the previous image is determined as a stopped object, there is a possibility of erroneous determination that actually determines a moving object as a stopped object. .

  SUMMARY An advantage of some aspects of the invention is that it provides an image processing apparatus capable of reducing erroneous determination.

  According to the first aspect of the present invention, an imaging unit that captures an image of the monitoring range, a reference image that is an image in which no moving object is present among images captured by the imaging unit, and an image captured by the imaging unit are sequentially reflected. Each of the intermediate images is stored, and the difference maintenance count maintained without changing the difference with respect to the reference image is stored for each pixel, and the reference image stored in the storage unit and the imaging unit are captured. A detection unit that detects the presence of a moving object based on a difference from the latest image, and the detection unit periodically and for each pixel, the difference between the pixel value in the latest image and the pixel value in the reference image. If the absolute value of the difference between the pixel value in the latest image and the pixel value in the reference image is smaller than the difference determination threshold, the moving object does not cover that pixel. Judge the difference maintenance count to 0 If the absolute value of the difference between the pixel value in the latest image and the pixel value in the reference image is larger than the difference determination threshold, the absolute value of the difference between the pixel value in the latest image and the pixel value in the intermediate image is further calculated. Compared to the difference determination threshold, if the absolute value of the difference between the pixel value in the latest image and the pixel value in the intermediate image is greater than the difference determination threshold, it is determined that the moving object is on that pixel and the number of difference maintenance times If the absolute value of the difference between the pixel value in the latest image and the pixel value in the intermediate image is smaller than the difference determination threshold, it is determined that the difference with respect to the reference image is maintained, and the difference maintenance count is set to 1. If the difference maintenance count reaches a predetermined stop determination threshold value of 2 or more, it is determined that a stop object different from the stop object existing in the reference image covers the pixel, and the pixel value in the reference image is determined as the intermediate image Update to the pixel value at 0 and set the difference maintenance count to 0 Together, despite the result of any comparison to reflect the pixel values of the latest image and updates the pixel value in the intermediate image.

  According to the present invention, since the reference image is updated only for the pixels in which the difference with respect to the intermediate image is maintained by the number of times of the stop determination threshold that is two or more, the stopped object can be obtained only by maintaining the difference with respect to the intermediate image once. It is possible to reduce erroneous determination that determines that a pixel that is not actually applied to the stopped object is applied to the stopped object, as compared with a case where it is determined that the image is applied to the pixel and the reference image is updated.

  The invention of claim 2 is characterized in that, in the invention of claim 1, the detection unit extracts and outputs a range of pixels having a difference maintenance count of 1 as a range of moving objects.

  The invention of claim 3 is characterized in that, in the invention of claim 1, the detection unit extracts and outputs a range of pixels having a difference maintenance count of 2 or more as a range of a stopped object.

  According to a fourth aspect of the present invention, in any one of the first to third aspects, the difference determination threshold is larger than a fluctuation range of the pixel value due to electromagnetic noise assumed in a use environment.

  According to a fifth aspect of the present invention, in any one of the first to fourth aspects, the imaging unit can change a gain, which is a parameter that determines a relationship between the incident light amount from the monitoring range and the pixel value, and detects the gain. When the gain is changed in the imaging unit, the unit updates the pixel value of each pixel of the reference image and the intermediate image and the difference determination threshold according to the changed gain.

  According to the present invention, it is possible to reduce erroneous determination due to a change in gain.

  According to a sixth aspect of the present invention, in any one of the first to fifth aspects, for each pixel, the pixel value Ep ′ in the intermediate image after the update by the detection unit is the intermediate image before the update by the detection unit. The pixel value Ep is represented by Ep ′ = αEn + (1−α) Ep using a predetermined positive number α of 1 or less and the pixel value En in the latest image.

  According to a seventh aspect of the present invention, in the second aspect of the present invention, the detection unit compares the number of pixels of each connected region with a predetermined noise determination threshold for each of the range of the moving object and the range other than the moving object, and moves Each pixel that is included in the range of the object but has the number of pixels that does not satisfy the predetermined noise determination threshold is excluded from the range of the moving object, and the difference maintenance count of each pixel included in the connected region is adjacent to the connected region. In addition to a connected area that is not included in the moving object range but the number of pixels is less than the noise determination threshold, the number of pixels that are included in the connected area The difference maintenance frequency is set to 1.

  According to the present invention, it is possible to reduce the influence caused by the part of the moving object where the pixel value coincides with the reference image or the stop object, or the influence caused by electromagnetic noise.

  In the invention of claim 8, in the invention of claim 3, the detection unit compares the number of pixels in each connected region with a predetermined noise determination threshold value for each of the range of the stopped object and the range other than the stopped object, and stops. A connected area whose number of pixels is included in the object range but does not satisfy the noise determination threshold is excluded from the range of the stopped object, and the difference maintenance count of each pixel included in the connected area is the difference between the pixels adjacent to the connected area. In addition to the number corresponding to the number of times of maintenance, a connected area that is not included in the stop object range but whose pixel count does not satisfy the noise determination threshold is added to the stop object range, and the difference between the pixels included in the connected area is maintained. The number of times is set to a number corresponding to the number of times of difference maintenance for each pixel adjacent to the connected region.

According to this invention, it is possible to reduce the influence caused by the part where the pixel value coincides with the reference image or the stop object in the moving object, or the influence caused by electromagnetic noise. The detection unit compares the number of pixels in each connected region with the same comparison result for each comparison with a predetermined noise determination threshold, and if the number of pixels is less than the noise determination threshold, each pixel belonging to the connected region The comparison result is reversed for each.

  According to the present invention, it is possible to reduce the influence caused by the part of the moving object where the pixel value coincides with the reference image or the stop object, or the influence caused by electromagnetic noise.

  According to a tenth aspect of the present invention, in the invention according to any one of the first to ninth aspects, the imaging unit is brought into the monitoring range by a light source that intermittently irradiates the monitoring range with light, an imaging element that images the monitoring range, and the light source. And a difference output unit that outputs a difference between the image when the light is irradiated and the image when the light is not irradiated to the monitoring range by the light source to the detection unit.

  According to the present invention, it is not affected by the brightness of the surrounding environment due to sunlight or lighting on / off.

  According to the first aspect of the present invention, the reference image is updated only for the pixels in which the difference with respect to the intermediate image is maintained by the number of stop determination thresholds equal to or greater than two, so the difference with respect to the intermediate image is only maintained once. Compared with the case where the reference image is updated by determining that the stopped object is applied to the pixel, the erroneous determination of determining that the pixel that is not actually applied to the stopped object is applied to the stopped object can be reduced. it can.

  In the invention of claim 5, when the gain is changed in the imaging unit, the pixel value of each pixel of the reference image and the intermediate image and the difference determination threshold are updated according to the changed gain. In addition, it is possible to reduce misjudgment caused by gain change.

  In each of the seventh to ninth aspects, since the determination result and the comparison result are inverted when the number of pixels in the connected region is less than the noise determination threshold value, the pixel value coincides with the reference image or the stop object by chance in the moving object. The influence by a part and the influence by electromagnetic noise can be reduced.

  According to a tenth aspect of the present invention, the imaging unit includes a light source that intermittently irradiates the monitoring range with light, an imaging element that images the monitoring range, an image when the monitoring range is irradiated with light, and a light source. Since the difference output unit that outputs the difference from the image when the monitoring range is not irradiated with light to the detection unit is output to the detection unit, an image in which the contribution of the ambient light to the pixel value is offset is output. Therefore, the influence of ambient light can be reduced.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  As shown in FIG. 2, the present embodiment includes an imaging unit B1 that has a solid-state imaging device such as a CCD image sensor or a CMOS image sensor and images a monitoring range, and an image that is captured by the imaging unit B1. And a storage unit B3 that stores a reference image in which no moving object exists, and a detection unit B2 that detects a moving object based on the difference between the latest image captured by the imaging unit B1 and the reference image stored in the storage unit B3. With.

  Next, the operation of the detection unit B2 of the present embodiment will be described mainly using FIG.

  In the storage unit B3, a reference image that is an image obtained by capturing an image of the monitoring range by the image capturing unit B1 in the absence of a moving object in advance, and an image (a new image obtained every time an image is obtained from the image capturing unit B1) Hereinafter, intermediate images that are sequentially updated to reflect “the latest image” are stored. The storage unit B3 stores a difference maintenance count N indicating the number of times that the difference with respect to the reference image is maintained without fluctuation for each pixel.

  As shown in FIG. 1, after the start (S1), the detection unit B2 periodically obtains an image from the imaging unit B1, and for each pixel of the obtained image (hereinafter referred to as “latest image”). The following operations are performed.

  First, the pixel value En in the latest image is acquired (S2). Next, the absolute value | En−Eb | of the difference between the pixel value En in the latest image and the pixel value Eb in the reference image is compared with a predetermined difference determination threshold Et (S3). If the absolute value | En−Eb | of the difference is smaller than the difference determination threshold Et, it is determined that the moving object does not cover the pixel, the difference maintenance count N is set to 0 (S4), and the pixel in the latest image is displayed. The pixel value Ep in the intermediate image is updated by reflecting the value (S5).

  On the other hand, if the absolute value | En−Ep | is larger than the difference determination threshold Et in step S3, the absolute value of the difference between the pixel value En in the latest image and the pixel value Ep in the intermediate image is further calculated as the difference determination threshold Et. (S6), if the absolute value | En−Ep | of the difference between the pixel value En in the latest image and the pixel value Ep in the intermediate image is larger than the difference determination threshold Et, the moving object is applied to the pixel. The difference maintenance count N is set to 1 (S7), and the process proceeds to step S5. If the absolute value | En-Ep | of the difference is smaller than the difference determination threshold Et in step S6, it is determined that the difference with respect to the reference image is maintained, and 1 is added to the difference maintenance count N ( S8), it is determined whether or not the difference maintenance count N has reached a predetermined stop determination threshold Nt (for example, Nt = 2) of 2 or more (S9). In step S9, if the difference maintenance count N has not reached the stop determination threshold value Nt, the process proceeds directly to step S5, and if the difference maintenance count N has reached the stop determination threshold value Nt, a stop object different from the stop object existing in the reference image. The pixel value Eb in the reference image is updated to the pixel value Ep in the intermediate image and the difference maintenance count N is set to 0 (S10), and the process proceeds to step S5. That is, the pixel value Ep is updated in the intermediate image in step S5 regardless of the comparison result between step S3 and step S6.

  Here, the pixel value Ep ′ in the intermediate image after the update in step S5 is the pixel value Ep in the intermediate image before the update by the detection unit B2, a predetermined positive number α of 1 or less, and the pixel value Ep ′ in the latest image. Using the pixel value En, it is expressed as Ep ′ = αEn + (1−α) Ep. α is a parameter that is appropriately selected according to the application. The closer α is to 1, the faster the change in the intermediate image is, and the faster the response to displacement, appearance / disappearance of the stationary object.

  After updating the pixel value Ep in the intermediate image in step S5, the process returns to step S2. At this time, if there is a pixel in the same latest image for which the above processing has not been completed, the same processing is performed for the next pixel in the same latest image, and if the above processing is completed for all the pixels, the next processing is performed. The same processing is performed in order from the first pixel in the latest image.

  According to the above configuration, since the reference image is updated only for the pixels in which the difference with respect to the intermediate image is maintained by the number of times of the stop determination threshold Nt that is a number equal to or greater than 2, the reference image is updated only once the difference with respect to the intermediate image is maintained. Compared with the case where the reference image is updated by determining that the object is applied to the pixel, it is possible to reduce erroneous determination that determines that the pixel that is not actually applied to the stopped object is applied to the stopped object.

  Here, as the stop determination threshold Nt is increased, the response to the displacement or appearance / disappearance of the stopped object is delayed, but an erroneous determination that actually determines that the object is not a stopped object is suppressed. The

  In selecting the difference determination threshold Et, for example, for an assumed moving object, the difference determination threshold Et is set larger than the fluctuation range of the pixel value within a period in which the pixel is on the moving object. If the difference determination threshold Et is selected as described above, the difference maintenance count N remains 1 during the period when the pixel is on the moving object, so that the reference image is erroneously updated within the period. Absent.

  When the above processing is completed for all the pixels of one latest image, before moving to the next latest image, only the range of pixels whose difference maintenance count N is 1 is extracted as the range of moving objects, or , By extracting only the range of pixels having the difference maintenance count N of 2 or more as the range of the stopped object, each connected region consisting of pixels included in the moving object or the stopped object and each connected region consisting of pixels not included In each case, the number of pixels constituting the connected area is compared with a predetermined noise determination threshold, and for connected areas where the number of pixels is less than the noise determination threshold, the attribute of being included or not included in the moving object or the stopped object is inverted. You may let them. In this case, for each pixel whose attribute has been inverted, the difference maintenance count is updated according to the difference maintenance count of each pixel adjacent to the connected region to which it belongs. For example, when pixels that have a difference maintenance count of 0 occupy more than half of the pixels adjacent to the connection area, the difference maintenance count of each pixel included in the connection area is set to 0, and the difference maintenance count is If less than half of the pixels are 0, the difference maintenance count of each pixel included in the connected area is set to the difference maintenance count of all the pixels whose difference maintenance count is not 0 among the pixels adjacent to the connected area. Average value (rounded down or rounded up). That is, when a connected area that is not included in the moving object is included in the moving object, the difference maintenance count of each pixel included in the connected area is 1. Here, there may be a part of the moving object where the pixel value coincides with the reference image or the stop object behind, and in such a case, the part of the moving object is not determined as a moving object or is stopped. An erroneous determination such as determination as an object may occur. Also, due to electromagnetic noise, misjudgment that a part of a moving object is not determined as a moving object or a part of a stopping object is not determined as a stopping object may occur. However, if the above configuration is adopted, the erroneous determination as described above can be reduced. The range of the moving object and the range of the stopped object extracted in the above process can be used for another process such as identification of the moving object and the stopped object in a later stage (not shown).

  Alternatively, instead of performing the series of processes from step S3 to step S8 consecutively for each pixel, the comparison between step S3 and step S6 is performed for all the pixels simultaneously, and each of step S3 and step S6 is performed. The comparison result is compared with the predetermined noise determination threshold value for the number of pixels constituting each of the connected regions that are composed of pixels having the same comparison result, and the comparison result is inverted for the connected region whose pixel number is less than the noise determination threshold value. You may make it perform a later process above. Here, inverting the comparison result means that the pixels belonging to the connected region whose absolute difference value is lower than the difference determination threshold value are treated as if the absolute value of the difference is higher than the difference determination threshold value. This means that the pixels belonging to the connected regions whose values are higher than the difference determination threshold are treated as those whose absolute difference value is lower than the difference determination threshold. By adopting this configuration, it is possible to reduce the influence caused by the part where the pixel value coincides with the reference image or the stop object accidentally in the moving object, or the influence caused by electromagnetic noise.

  Further, when the gain is changed in the imaging unit B1, the detection unit B2 determines each pixel value of the reference image and the intermediate image stored in the storage unit B3, the difference determination threshold value, and the difference according to the changed gain. May be changed respectively. That is, the larger the gain (that is, the larger the pixel value with respect to the incident light amount), the larger the respective pixel values of the reference image and the intermediate image and the difference determination threshold value. By adopting this configuration, it is possible to reduce erroneous determination due to gain change.

  Here, as shown in FIG. 3, as the imaging unit B <b> 1, the light source 2 that irradiates light to the monitoring range and the reflected light from the object 5 existing in the monitoring range are incident via the optical system 6. An image sensor 10 as an image sensor, an image generation unit 3 that generates an image using the output of the image sensor 10, and a control unit 4 that controls the light source 2 and the image sensor 10 can be used. .

  The image sensor 10 is a CCD image sensor including a plurality of light receiving elements 1 arranged in a matrix, for example. Each light receiving element 1 includes a photosensitive portion 1a that generates charges according to the amount of incident light and a charge storage portion 1b that stores charges generated by the photosensitive portion 1a. The light receiving element 1 is a photoelectric conversion element such as a MIS (Metal-Insulate-Semiconductor) element. Each light receiving element 1 has a control electrode 13, and the control unit 4 controls the voltage applied to the control electrode 13, thereby controlling the magnification, that is, gain, of the charge (pixel value) output with respect to the incident light amount. To do. Further, the image sensor 10 includes a charge extraction unit 1 c that individually extracts the charges accumulated in the charge accumulation unit 1 b of each light receiving element 1 and transfers them to the image generation unit 3. That is, an amount of charge corresponding to the amount of light incident on the photosensitive portion 1a is accumulated in the charge accumulation portion 1b, and this amount of charge is output to the image generation portion 3 as a pixel value.

  As the light source 2, for example, a light emitting diode can be used. The light source 2 is controlled so as to be intermittently lit by the control unit 4. As shown in FIG. 4, the image generation unit 3 receives the pixel value Aa when the light source 2 is turned on and the pixel value Ab when the light source 2 is turned off, and the pixel value Aa when the light source 2 is turned on and the light source 2 is turned off. An amplitude image having a pixel value that is the difference Aa−Ab from the pixel value Ab is generated and output to the detection unit B2. That is, since the contribution of the ambient light to the pixel value as indicated by the curve E in FIG. 4 is offset by taking the difference Aa−Ab as described above, the amplitude image is less affected by the fluctuation of the ambient light. An object that is closer to the imaging unit B1 and is more strongly illuminated by the light source 2 has an enhanced image. For example, the duty ratio of lighting of the light source 2 is 50% (that is, the lighting time and the light-off time are the same). It is desirable that the blinking frequency of the light source 2 is set to a period high enough that blinking cannot be confirmed by human eyes, for example, 10-100 kHz. Here, when only one of the pixel value Aa when the photoelectric 2 is turned on and the pixel value Ab when the light is turned off is used for detection of a moving object, a change in the image due to a change in environmental light such as sunlight or streetlight light. However, it may be mistaken for an image change caused by a moving object and may cause erroneous detection. On the other hand, if the pixel value Aa-Ab of the amplitude image is used for detection of a moving object as described above, erroneous detection due to fluctuations in ambient light is reduced. Note that the pixel value of the amplitude image does not necessarily have to be calculated for each blinking. For each of the plurality of blinks, the total of the plurality of pixel values Aa that are turned on and the plurality of pixel values Ab that are turned off. It may be calculated as a difference from the total of the batches.

  In addition to the amplitude image generation unit 3a serving as a difference output unit that generates an amplitude image as described above, the image generation unit 3 generates a grayscale image generated using only the pixel value Aa when the light source 2 is lit. A gray-scale image generating unit 3b. The grayscale image is close to the image seen with the naked eye, and is used for display on a display such as a liquid crystal screen.

It is a flowchart which shows operation | movement of the detection part in embodiment of this invention. It is a block diagram which shows schematic structure same as the above. It is a block diagram which shows schematic structure of an example of an imaging part same as the above. It is explanatory drawing which shows operation | movement of the imaging part of FIG. An example of an image captured by the imaging unit is shown, (a) shows an example of an image without a moving object, (b) shows an example of an image with a moving object, and (c) shows an example from (b). The example of the image from which the moving object was extracted is shown.

Explanation of symbols

2 Light source 3a Amplitude image generation unit (difference output unit in claims)
10 Image sensor (Image sensor in claims)
B1 imaging unit B2 detection unit B3 storage unit

Claims (10)

An imaging unit for imaging the monitoring range;
A reference image, which is an image captured by the imaging unit, in which there is no moving object, and an intermediate image in which images captured by the imaging unit are sequentially reflected are stored, and a difference with respect to the reference image is changed. A storage unit for storing the number of difference maintenance times maintained for each pixel;
A detection unit that detects the presence of a moving object based on the difference between the reference image stored in the storage unit and the latest image captured by the imaging unit;
The detection unit periodically and for each pixel compares the absolute value of the difference between the pixel value in the latest image and the pixel value in the reference image with a predetermined difference determination threshold, and the pixel value in the latest image and the reference image If the absolute value of the difference from the pixel value at is smaller than the difference determination threshold, it is determined that the moving object does not cover that pixel, the difference maintenance count is set to 0, and the pixel value in the latest image and the pixel in the reference image If the absolute value of the difference from the value is larger than the difference determination threshold, the absolute value of the difference between the pixel value in the latest image and the pixel value in the intermediate image is compared with the difference determination threshold, and the pixel value in the latest image If the absolute value of the difference between the pixel value in the intermediate image and the pixel value in the intermediate image is larger than the difference determination threshold, it is determined that the moving object is on that pixel, the difference maintenance count is set to 1, and the pixel value in the latest image and the intermediate image If the absolute value of the difference from the pixel value is smaller than the difference judgment threshold, It is determined that the difference is maintained, 1 is added to the difference maintenance count, and if the difference maintenance count reaches a predetermined stop determination threshold value of 2 or more, a stop object different from the stop object existing in the reference image is the pixel. The pixel value in the reference image is updated to the pixel value in the intermediate image and the difference maintenance count is set to 0, and the pixel value in the latest image is changed regardless of the result of any comparison. An image processing apparatus that updates a pixel value in an intermediate image by reflecting.   The image processing apparatus according to claim 1, wherein the detection unit extracts and outputs a range of pixels having a difference maintenance count of 1 as a range of moving objects.   The image processing apparatus according to claim 1, wherein the detection unit extracts and outputs a range of pixels having a difference maintenance count of 2 or more as a range of a stopped object.   The image processing apparatus according to claim 1, wherein the difference determination threshold is larger than a fluctuation range of a pixel value due to electromagnetic noise assumed in a use environment. The imaging unit can change a gain that is a parameter for determining the relationship between the incident light amount from the monitoring range and the pixel value,
When the gain is changed in the imaging unit, the detection unit updates the pixel value of each pixel of the reference image and the intermediate image and the difference determination threshold according to the changed gain. The image processing apparatus according to claim 1.   For each pixel, the pixel value Ep ′ in the intermediate image after the update by the detection unit is the pixel value Ep in the intermediate image before the update by the detection unit, a predetermined positive number α of 1 or less, and the latest image The image processing apparatus according to claim 1, wherein the pixel value En is expressed by Ep ′ = αEn + (1−α) Ep.   The detection unit compares the number of pixels in each connected region with a predetermined noise determination threshold for each of the range of the moving object and the range other than the moving object, and the number of pixels included in the range of the moving object is the predetermined noise. A connected area that does not satisfy the determination threshold is removed from the range of the moving object, and the number of times of difference maintenance for each pixel included in the connected area is set to a number corresponding to the number of times of difference maintenance for each pixel adjacent to the connected area. A connected area that is not included in the range but whose number of pixels does not satisfy the noise determination threshold is added to the range of the moving object, and the difference maintenance count of each pixel included in the connected area is set to 1. 2. The image processing apparatus according to 2.   The detection unit compares the number of pixels in each connected region with a predetermined noise determination threshold value for each of the range of the stopped object and the range other than the stopped object, and the number of pixels is included in the range of the stopped object while the number of pixels is the noise determination threshold value. And the number of times of maintaining the difference of each pixel included in the connected region is determined according to the number of times of maintaining the difference between the pixels adjacent to the connected region. The number of pixels that are not included in the connected area but the number of pixels is less than the noise determination threshold is added to the range of the stopped object, and the number of times of maintaining the difference of each pixel included in the connected area is the number of times of maintaining the difference between the adjacent pixels. The image processing apparatus according to claim 3, wherein the number is set in accordance with.   For each comparison, the detection unit compares the number of pixels in each connected region where the comparison results match each other with a predetermined noise determination threshold value. If the number of pixels is less than the noise determination threshold value, the detection unit relates to each pixel belonging to the connected region. The image processing apparatus according to claim 1, wherein each of the comparison results is inverted. The imaging unit
A light source that intermittently irradiates the monitoring range with light;
An image sensor for imaging the monitoring range;
A difference output unit that outputs a difference between an image when the monitoring range is illuminated by the light source and an image when the monitoring range is not illuminated by the light source to the detection unit The image processing apparatus according to claim 1.

Images