JPH10283466A - Device for detecting intruding object - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an intruding object detector for surely detecting an intruding object to be detected even when the wide range of a water surface whose background state is unstable on the sea or the like is an area to be monitored. SOLUTION: Picture data image-picked up by a night vision camera 1 are quantized into 256 gradations by an A/D converting part 11, the obtained picture data are blocked into 10×10 local areas by a blocking processing part 13a, and the luminance average of each block is calculated. Then, when the state of high luminance is continued in more than a contact period, it is judged as a fixed object such as an island by a mask judgment value calculating part 13b and a mask processing part 13c, and the pertinent part is masked. A threshold value for detecting a ship being an intruding object is set based on the luminance of a water surface by an object judgment value calculating part 13d, and when the state of luminance beyond the threshold value is continued in more than a constant period, it is judged as a ship by an intruding object judgment processing part 13e, and a detection signal is outputted, and an alarm 20 is operated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an intruder detecting apparatus, and more particularly to an apparatus for monitoring an intruder on a water surface such as a sea, a river or a lake using a night vision camera.

[0002]

BACKGROUND OF THE INVENTION As an automatic monitoring system using a camera (night-vision camera) that captures heat rays or infrared rays generated from an object, an automatic monitoring system detects an object entering or existing in a monitoring area, and detects a type of the detected object. Various methods have been proposed and put into practical use, such as recognizing and tracking a detected object when it moves. However, those that have been put into practical use have been limited to the land where the background portion of the monitoring area does not change much.

[0003] That is, when the monitoring area is on the water surface,
Since the state of the water surface is constantly changing due to waves generated on the water surface, even if the terrestrial system is applied as it is, many malfunctions cannot be provided for practical use. There is also a monitor area provided on the screen to increase the stability by calculating a change in luminance when an object blocks the monitor area by an integration process. There is a difference, and when the monitoring area is at sea, the monitoring position also changes due to the rise and fall of the tide, so that it is necessary to follow the change. Therefore, the setting must be constantly changed and adjusted. Furthermore, in the case of water surface monitoring (especially sea monitoring), the monitoring area is wide and one camera must be turned at regular intervals to monitor the wide monitoring area in order. Therefore, the above setting change / adjustment must be performed each time the camera is turned. Therefore, it is practically difficult to perform such changes and adjustments sequentially, and it has not been possible to automatically monitor the water surface in the past.

[0004] The present invention has been made in view of the above background, and an object of the present invention is to solve the above-mentioned problem, and to provide a monitoring area on a water surface having an unstable background such as the sea, and furthermore, a monitoring area. It is an object of the present invention to provide an intruder detecting device capable of reliably detecting an object (intruder) to be detected even if the object is a wide area.

[0005]

In order to achieve the above object, an intruder detecting apparatus according to the present invention includes a camera for imaging heat rays or infrared rays generated from an object. It is assumed that an intruding object detection device detects a intruding object having a water surface and entering the imaging range of the camera. Then, a luminance calculating unit (corresponding to the “blocking processing unit 13a” in the embodiment) that calculates the luminance of each unit within the detection range in the image data captured by the camera, and the luminance calculating unit calculates the luminance. Brightness change information extracting means for obtaining brightness change information of each part based on the brightness,
Detecting means for discriminating and detecting the intruder to be detected from the water surface based on the luminance change information obtained by the luminance change information extracting means (in the embodiment, the luminance change information extracting means and the detecting means (Integrated in the object determination processing unit 13e)). Here, the “each unit” may be a block unit including a predetermined local area as described in the embodiment, or may be a single pixel unit. In the case of a block unit, the image is blurred, so that the image becomes blurred. In the case of an object whose position constantly changes in a small area such as a wave,
The flicker is very short, so it is easy to eliminate.

In the present invention, the water surface has a low luminance when photographed by a camera defined in the claims. On the other hand, the brightness of an island, a building, a ship, a bird, a wave, or the like that can exist within the imaging range of the camera near or above the water surface increases. Therefore, the water surface and the rest can be separated from the magnitude relationship of the luminance. That is, when looking at the luminance change information, the part where the low luminance state continues can be determined to be the water surface. In the case of a fixed object such as an island even in a portion where the luminance is high, since the state of high luminance continues, it can be determined from the luminance change information that the object does not move, that is, it is not an intruder. Further, since a ship, a bird, a wave, and the like have different sizes and different moving speeds, the brightness change state (information) focusing on one certain portion is different. Thus, from such luminance change information, it can be determined whether or not a portion having a high luminance at a certain time is an intruder to be finally detected.

On the premise of the above configuration, a threshold value (corresponding to the “object determination value” in the embodiment) corresponding to the luminance of the water surface portion is set, and the intruder is used by using the threshold value. May be detected (claim 2). The object for automatically setting the threshold can be realized by the object determination value calculator 13d described in the embodiment.

Further, prior to the process of detecting an intruding object, a portion having a characteristic in the luminance change information (in the embodiment, the luminance of an island or the like is high, but is fixed and has a small change in luminance, and is not an intruding object to be detected. A mask processing unit (in the embodiment, realized by the “mask determination value calculation unit 13b and the mask processing unit 13c”) that performs a mask process on the portion may be configured (claim 3).

With this configuration, the masked portion is not searched in the subsequent intruding object detection processing, so that the time required for the detection processing can be shortened and the detection can be performed with high accuracy.

A specific configuration and algorithm of the detection means is, for example, that the intruder is determined to be an intruder when a period in which a state of a predetermined luminance or higher at the same point in the detection range is continuous meets a predetermined condition. (Claim 4), whether the intruder is the intruder based on the size of an area where a portion having a predetermined brightness or more is continuous (Claim 5), or a combination of them as appropriate. it can.

[0011]

FIG. 1 shows a preferred embodiment of an intruder detecting apparatus according to the present invention. In the present embodiment, the present invention is applied to an intruding object detection device incorporated in an automatic monitoring system for detecting a ship that has entered the sea in a monitoring area where the water surface to be monitored is at sea.

As shown in FIG. 1, an output (video signal) of the night-vision camera 1 for picking up heat rays or infrared rays is given to an A / D converter 11 in an image processing unit 10 where a predetermined sampling time is applied. The video signal (luminance) is digitized and quantized with 8 bits (256 gradations) suitable for image processing. Then, the digitized image data (for one frame) is stored in the next-stage image memory 12.
Is stored in

[0013] As a result, the luminance of the portion that emits heat is increased, and the luminance of the pixel of the portion where a ship as a detection target or an island, a building, a bird, a wave, or the like as a non-detection target exists is reduced. Get higher. In the present embodiment, the description will be made on the assumption that the higher the luminance, the larger the numerical value. Therefore, when white with high luminance is “0” and black with low luminance is “255”, the magnitude relation (Yes / N
It goes without saying that the positions are reversed at appropriate places, such as o) and the positive / negative of a constant when obtaining the determination value.

The image data stored in the image memory 12 is subjected to predetermined image processing by an image processing section 13, and an object to be detected (an intruder) is included in the captured image data in the monitoring area. It is determined whether or not it exists. Specifically, the input side includes a blocking processing unit 13a, which blocks the entire captured image data into a local area of 10 × 10 pixels, and calculates an average luminance for each block. This luminance average is obtained by simply dividing the sum of the luminance of each pixel constituting one block by the number of pixels (100).

A mask determination value calculation unit 13b is provided downstream of the block processing unit 13a. The mask determination value calculation unit 13b has a function of performing a predetermined process according to the flowchart shown in FIG.

First, the luminance average of the blocks in the entire area obtained by the block processing unit 13a is obtained, and based on the average, the average luminance of the entire area (overall average luminance) and its deviation (overall luminance deviation: σ) are obtained. It is calculated (ST1, ST2). Then
The mask determination value f1 (σ) is obtained by substituting into the following equation (1) (ST3).

[0017]

F1 (σ) = overall average luminance + constant × σ (1) Here, an arbitrary value among 10 to 30 is set as the constant. As a setting criterion at this time, the smaller the constant, the smaller the margin with respect to the overall average luminance, so that it responds even to an object with a small luminance. I do. However, if the value is set to a small value, many non-detection target objects are also detected, so that the determination process takes time and the possibility of erroneous detection increases.

Therefore, for example, when the wave is gentle, since the change in the water surface is relatively small, the constant is set small and the object is reliably detected by reacting to even a slight change in luminance. . Conversely, when the waves are rough, even when the moving surface such as the detection target does not invade relatively much changes in the water surface, the brightness changes in blocks are repeatedly performed due to the influence of the waves. It is preferable to set a large value to make it difficult to detect the wave itself (having lower luminance than the object to be detected) and to prevent malfunction.

Further, in the present embodiment, it is determined whether or not the mask determination value f1 (σ) obtained as described above is greater than or equal to a predetermined minimum sensitivity value (ST4).
In this case, the obtained mask determination value f1 (σ) is determined as the final mask determination value f1 (σ) and is output. Also,
In the case of “not more than the minimum sensitivity value”, the minimum sensitivity value is output as the mask determination value f1 (σ) (ST5).

Thus, for example, when the number of heat-generating objects is small and the brightness of the water surface existing in the monitoring area is small overall, the mask determination value f1 (σ) may be extremely small. If it is determined based on a small value whether it is a portion to be masked as described later or not and the corresponding portion is masked, the mask is erroneously applied even though it is originally a water surface, and it becomes an undetected area. There is a possibility that an object that has entered the mask portion may not be detected. However, by setting the minimum sensitivity value as in the present embodiment, occurrence of such a situation can be suppressed as much as possible.

The mask determination value f1 thus obtained
(Σ) is sent to the mask processing unit 13c at the next stage, where a non-detection target object such as an island or a building that is not moving and that does not move is extracted and masked, and the subsequent object detection processing is performed. In this case, the mask area is not detected / searched to prevent erroneous detection and to shorten the processing time. The mask processing unit 13c has a function of executing a flowchart as shown in FIG.

That is, first, the mask judgment value calculation unit 13
The mask determination value f1 (σ) output from b is obtained (ST11). Then, the luminance average and the mask determination value f1 of each block blocked by the blocking processing unit 13a are obtained.
(Σ) are compared (ST12). When the average luminance of the block is smaller than the mask determination value f1 (σ),
Since the block is on the water surface or the like, the process proceeds to step 13 and shifts to processing of another block. Step 1
If the branch determination in step 2 is equal to or greater than the mask determination value f1 (σ), there is a possibility that a heating element such as an island is present.
And successively detect the corresponding number of times. Blocks having an average luminance equal to or greater than the mask determination value f1 (σ) consecutively at least n times as the set value generate predetermined heat and do not move. Since it can be estimated to be a background portion other than the water surface, the corresponding block is masked (ST15, ST16). In addition, even if the block is determined to be Yes in the branch determination in step 12 and proceeds to step 14, the result of the determination is that the moving object such as a ship happens to move in the area corresponding to the block. In the meantime, there is no moving object in the area after a certain period of time, so by setting the set number n to an appropriate value, only fixed objects such as islands can be extracted and masked.

That is, the relationship between the luminance change per block and the time when an island, a ship, a bird, and a wave on the water surface are picked up by a night-vision camera is shown in FIGS. 4A to 4D. Become. As shown in FIG. 2A, in the case of a fixed object such as an island, the brightness obtained by imaging with the night vision camera 1 is equal to or more than the determination value f1 (σ) because it generates heat, and is constant. The value is held (there is no change in luminance). Therefore, the time t1 at which f1 (σ) or more is continuous is infinite in principle.

In the case of a ship, as shown in FIG. 3B, when the ship is present in the area corresponding to the block, the brightness is equal to or higher than the set value. And the set value f1 only during a certain time t2.
(Σ) or more, and when there are no other ships, the brightness of the water surface is low (low), so it is less than the set value. At this time, since the ship is relatively large, it takes a relatively long time for the ship from the bow to the stern to pass through one block, and the moving speed is slow, so that the time t2 is relatively long. The specific length differs depending on the moving speed and size of the ship at that time.

On the other hand, since the bird is moving also in the case of a bird, the brightness of the block is set to the set value f1 (σ) for a certain time t3.
That is all. Since the bird is smaller than the ship and has a higher moving speed, the moving time t3 of the bird is much shorter than the time t2 when the ship passes.

Similarly, in the case of a wave, the luminance of the block becomes equal to or more than the set value f1 (σ) for a certain time t4. In addition, in the case of a wave, a phenomenon in which the luminance becomes large appears periodically. Since the waves are also smaller and the moving speed is faster than that of the ship, the time t4 is much shorter than the time t2 when the ship passes.

Since the times t3 and t4 are often several hundreds ms or less, if the sampling time at the time of A / D conversion is taken in at an interval of several hundred ms, the block of the block may be captured by birds or waves. Brightness average is set value f1
(Σ) will not be exceeded or, if at all, it will be one single shot. Then, by appropriately setting the set value n so as to be equal to or longer than the time t2 required for the ship to pass, only the blocks in which the fixed object that generates heat, such as an island, is imaged are determined in step 15 in FIG. Yes
And a mask can be applied.

The mask processing shown in step 16 is as follows.
A mask is literally applied to the corresponding block, and the process is performed only on an area (block) other than the mask as an undetected area in the subsequent object detection processing, thereby shortening the processing and preventing erroneous detection. It has become. In the present embodiment, the masked area is displayed on the monitor. Therefore, the mask areas are grouped according to the state of the adjacent blocks, and only the outline is displayed. Specifically, for example, when a predetermined number of mask areas are present as shown in FIG. 5A, they are integrated and only the outline is grouped and displayed as shown in FIG. 5B. I have to. Thereby, a large number of frames (4 of each block) are displayed on the monitor.
Side) is displayed and is hard to see.

Then, the mask processing unit 13c stores the position information of the mask area extracted as described above, and in the actual monitoring (object detection), the mask processing unit 13c thereafter blocks the mask area by the blocking processing unit 13a. After performing the mask processing on the image data that has been masked, the masked image data is provided to the next-stage object determination value calculation unit 13d.

The object determination calculation section 13d has a function of performing predetermined processing according to the flowchart shown in FIG. That is, first, image data that has been blocked and masked is obtained (ST21). And
Based on this, the average luminance (overall average luminance) of the unmasked area and its deviation (overall luminance deviation: σ) are calculated, and then substituted into the following equation (2) to determine the object determination value f2.
(Σ) is obtained (ST22).

[0031]

F2 (σ) = overall average luminance + constant × σ (2) Here, the constant is set to any value from 10 to 30. The setting criterion at this time is basically the same as the above-described calculation of the mask determination value. The smaller the constant, the higher the sensitivity. Therefore, an appropriate value is set according to the state of the water surface. The specific constant is the mask determination value f
It may be the same as the constant used for obtaining 1 (σ) or may be different.

Further, in the present embodiment, it is determined whether or not the object determination value f2 (σ) obtained as described above is greater than or equal to a predetermined minimum sensitivity value (ST23), and “more than the minimum sensitivity value” is determined.
In the case of, the obtained object determination value f2 (σ) is determined and output as the final object determination value f2 (σ). In the case of “not greater than or equal to the minimum sensitivity value”, the minimum sensitivity value is set as the object determination value f2 (σ), and the next intruder determination processing unit 13 is used.
e (ST24, ST25). As a result, when there is no object to be detected, the object determination value f2 (σ) becomes an extremely small value, and a portion having a high luminance on the water surface is not erroneously detected as an object.

The intruder determination processing section 13e has a function of executing a flowchart as shown in FIG. That is, first, the object determination value f2 (σ) output from the object determination value calculation unit 13d is obtained (ST31). And
The average luminance and the object determination value f2 of each block (excluding the mask area) blocked by the blocking processing unit 13a
(Σ) are compared (ST32). If the average luminance of the block is smaller than the object determination value f2 (σ), the block is on the water surface or the like, and the process proceeds to step 33 and proceeds to processing of another block. If the branch determination in step 32 is equal to or greater than the object determination value f2 (σ), there is a possibility that the detection target object (ship) is moving, so the process jumps to step 34 to detect the corresponding number of times continuously. . Then, the object determination value f2 (σ) is continuously set value m or more (usually smaller than the set value n at the time of extracting the mask described above) times or more.
Blocks having the above average luminance generate predetermined heat, and have a somewhat large shape of the object and are moving slowly (an object that does not move is not detected here because it is masked). ), It is estimated that the ship is a detection target object, and it is determined that an object is detected (ST).
35, 36). Even in the case of a block that was determined to be Yes in the branch determination of step 32 and proceeded to step 34, the result of the determination is that a moving object such as a bird or a wave or the like accidentally moved the area corresponding to the block. In the case of the cause, the moving object disappears in the area after a very short time has elapsed, so that only the ship can be detected by setting the set number m to an appropriate value.

This principle is clear from the relationship between the luminance and the time shown in FIG. However, in the case of this object detection, m is set so as to be longer than the times t3 and 4 and shorter than the time t2, and in order to reliably detect the ship, the moving time is relatively long in the ship. It is necessary to set the time to be shorter than t2 when assuming that the object moves fast and the shape is small, and t2 (moving speed is slow and the shape is large) is assumed in the mask determination process. Shorter than

As described above, a ship to be detected can be accurately detected in accordance with a change in luminance of a certain block.

More preferably, the state of adjacency of the block determined as the object to be detected thus determined is determined, grouping processing is performed in the same manner as mask processing, and the size of the area grouped by the grouping is reduced. ,
If the size is equal to or larger than the set size or within a certain range, the ship may be determined as the detection target. In this case, the determination can be made with higher accuracy.

When an intruding object to be detected is detected in this way, a detection output is sent to the control unit 14. Then, the control unit 14 operates the alarm 20 based on the detection output to output an alarm. As the alarm, various alarms such as a sound such as a buzzer or a visual alarm such as a warning light or other lamp can be used.

[0038] Regardless of the presence or absence of object detection, image data picked up by the night-vision camera 1 and various image information indicating a mask area and an intruding object are temporarily stored in the image memory 15 and D / A converted. A monitor is output to the monitor TV 21 via the unit 16 so that a monitoring person can perform visual monitoring.

Next, while explaining the operation of the above-described device, auxiliary functions that have not been described in the above-described embodiment will also be described. For example, when an image is taken by a night vision camera, the island 30 is brighter and brighter than the other water surface 31 as shown in FIG. Similarly, as shown in FIG. 9, the ship 32 also appears brighter and brighter than other water surfaces 31 and the like. Further, as shown in FIG.
3, and the wave 34 is also bright and has high brightness as shown in FIG.

Therefore, before shifting to the actual monitoring processing, the monitoring area is actually imaged and a mask area is set.
That is, an image obtained by converting image data captured by the night-vision camera 1 into digital data of 256 gradations via the A / D converter 11 is stored in the image memory 12. The storage of the image data is sequentially executed at a fixed sampling time (for example, at intervals of several hundred ms). Then, the stored image data is supplied to the block processing unit 13a to block the image data into a local area of 10 × 10 pixels. Thereby, for example, image data as shown in FIG. 12 is obtained.

Next, the mask decision value is calculated by the mask decision value calculating unit 13b for the sequentially applied image data to obtain the mask decision value, and the mask processing unit 13c performs the flowchart shown in FIG. 3 based on the mask decision value. A state of high luminance equal to or greater than the mask determination value is continuously n
A mask area is set by extracting blocks that have been repeated successively and performing grouping. Then, in the present embodiment, at the time of the initial setting until the setting of the mask region, the substantial operation is not performed after the mask processing unit 13c. Therefore, the masked image data is stored in the image memory 14 as it passes through each part as it is, and finally the monitor TV 2
1 is output to the monitor. In the display example at this time, for example, a mask area in which a predetermined portion in a line segment (open) indicating the monitoring area R in image data captured as shown in FIG. 13 is surrounded by a frame of a predetermined size MR is set. In the present embodiment, since the grouping process is performed on the mask region MR, only the outer frame portion of the mask region MR is displayed as a white frame (see FIG. 14). Thus, the mask processing ends. The timing of the end may be such that the mask processing is automatically ended when the image is acquired a set value n times or a certain number of times larger than the set value, or the investigator manually ends the processing. A command may be input.

Next, the mode shifts to the actual monitoring mode. After shifting to the monitoring mode in this way, the mask determination value calculation unit 13b enters the through state, and the mask processing unit 13c does not detect the mask area but extracts the mask area by the previously performed mask processing. A mask MR is applied to a portion corresponding to the mask region (FIG. 14). That is, after blocking the image data stored in the image memory 12 by the blocking processing unit 13a, the mask processing unit 13c masks a predetermined block.

Then, based on the luminance information of each block in the unmasked portion, the object judgment value calculation unit 13d extracts a block that satisfies a predetermined condition (continues to exceed the judgment value continuously for a set number of times m), The detection output of the moving object (intruder 36) is performed (FIG. 15). In this embodiment, since the grouping is performed, one intruder is indicated by one frame. The alarm display 37 is displayed on the monitor TV 21 in an overlapping manner.

16A and 16B, it is confirmed that the flying bird 33 is not extracted and only the relatively large and slowly moving ship 32 is detected. Was.

The timing of switching between the above-described mask processing and the detection mode is as follows. For example, when the area to be imaged by the night vision camera 1 is fixed (the camera does not turn), the mask processing is performed at regular intervals and the mask area is changed. By performing the resetting, it is possible to follow up and cope with, for example, a difference in luminance of a water surface or the like between day and night and a change in the water surface position due to ebb and flow of the tide. On the other hand, in the case where the night vision camera is turned to monitor a wide area, a mask area may be set by performing mask processing every time the camera is turned and stopped at a predetermined position. Thus, even if the background image of the monitoring area changes each time the vehicle turns, a fixed object such as an island can be detected and masked automatically in accordance with the change.

Further, in this embodiment, the mask area can be set by manual operation. That is, FIG.
As shown in (A), a boundary line 38 that partitions each block is displayed on the screen of the monitor TV 21. At this time, the image is superimposed on the captured image and displayed. Then, the pointer P can be moved as shown in FIG. 3B to indicate a region to be masked and deleted from the detection region. This allows
For example, it is possible to manually remove an area such as in the air where it is clear that a ship does not exist before performing the automatic mask processing, thereby performing more accurate automatic mask processing. This method works particularly well for systems that do not rotate the camera.

[0047]

As described above, the intruding object detecting apparatus according to the present invention can detect the type of the object in the high luminance portion based on the luminance change, so that the background state such as the sea is unstable. The water surface serves as a monitoring area, and even if the monitoring area is wide, an object (intruder) to be detected can be reliably detected.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a preferred embodiment of an intruder detecting device according to the present invention.

FIG. 2 is a flowchart illustrating functions of a mask determination value calculation unit.

FIG. 3 is a flowchart illustrating functions of a mask processing unit.

FIG. 4 is a diagram illustrating a characteristic of a time-luminance change with respect to a type of an object for explaining an operation principle of a mask processing unit and an intruding object determination processing unit;

FIG. 5 is a diagram illustrating functions of a mask processing unit.

FIG. 6 is a flowchart illustrating a function of an object determination value calculation unit.

FIG. 7 is a flowchart illustrating functions of an intruder determination processing unit.

FIG. 8 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 9 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 10 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 11 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 12 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 13 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 14 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 15 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 16 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

FIG. 17 is a diagram showing a display example of an actual monitor screen for explaining the operation of the present invention.

[Explanation of symbols]

 Reference Signs List 1 night vision camera 10 image processing unit 12 image memory 13 image processing unit 13a blocking processing unit 13b mask determination value calculation unit 13c mask processing unit 13d object determination value calculation unit 13e intruding object determination processing unit

Claims (5)

[Claims] An intruding object detection device, comprising: a camera for imaging heat rays or infrared rays generated from an object, wherein an imaging range of the camera includes a water surface, and an intruding object that enters the imaging range of the camera is detected. A luminance calculating means for calculating the luminance of each part within a detection range in the image data captured by the camera; and a luminance change information extracting means for obtaining luminance change information of each part based on the luminance calculated by the luminance calculating means. And an intruder detecting means for distinguishing and detecting the intruder to be detected from a water surface portion based on the luminance change information obtained by the luminance change information extracting means. 2. The intruder detection device according to claim 1, wherein a threshold value is set according to the brightness of the water surface portion, and the intruder is detected using the threshold value. . 3. The intruder detection apparatus according to claim 1, further comprising a mask processing means for masking a portion having a characteristic in the luminance change information before the detection processing of the intruder. 4. The apparatus according to claim 1, wherein the detecting means determines that the object is the intruder when a period during which a state having a predetermined luminance or higher at the same position in the detection range is continuous meets a predetermined condition. Item 4. The intruder detection device according to any one of Items 1 to 3. 5. The apparatus according to claim 1, wherein said detecting means determines whether or not the object is the intruder based on the size of a region where a portion having a predetermined luminance or higher is continuous. 2. The intruder detection device according to claim 1.