JP6412032B2 - Space recognition device - Google Patents

Description

  The present invention relates to a space recognition device that recognizes a change in a space from a measurement signal (for example, luminance signal) obtained by continuously measuring a physical quantity (for example, luminance) in a predetermined space.

  For the purpose of crime prevention or the like, it has been conventionally performed to detect a change area by an intruder or the like by performing background difference processing on an image captured by a surveillance camera. In detecting such a change area, erroneous detection due to shaking of a plant or illumination variation has been a problem.

  Conventionally, there is a case where a technique called a background model is adopted in order to reduce false detection due to shaking of a plant or illumination variation.

  For example, in the intruding object detection device described in Patent Literature 1, a background model that represents an appearance probability distribution of pixel values in each pixel of a plurality of monitoring images is generated, and the newly captured monitoring image is compared with the background model. In addition, it is described that a change area by an intruder is detected by obtaining a correlation value in each pixel.

JP 2011-123742 A

  However, in the prior art, fluctuations on the background image information side are reduced, but the influence of fluctuations on the side compared with the background image information (the monitoring image side in which the intruding object is reflected in the above description) still remains. In addition, the detection accuracy of the change region is reduced due to these fluctuations.

  In other words, in a normally used monitoring camera, each image sensor outputs an average value of luminance signals for a fixed time (for example, 1/5 second) as the luminance value of the pixel corresponding to the image sensor. The luminance value output in this way is unstable because it always fluctuates due to the disturbance generated within the predetermined time and the noise inside the camera. Therefore, as long as the change area is detected using this luminance value, it has been difficult to avoid the occurrence of an error due to fluctuation such as a part of the change area missing or a background mixed in a part of the change area.

  In addition, the correlation value calculated for detecting the change region in the prior art is a result obtained by calculating a plurality of luminance values including fluctuation, and is a scalar. In the correlation value, which is the calculation result of multiple luminance values including fluctuations, the influence of fluctuation can be doubled, and since it is a scalar, non-fluctuation components are separated from the calculation results and errors due to fluctuation components are eliminated. It was difficult to do. Although there is an example in which a difference value is calculated instead of calculating a correlation value, since the influence of fluctuation can be doubled similarly to the correlation value, the non-fluctuation component is difficult to separate.

  Therefore, the present invention has been made in view of the above problems, and provides a space recognition device that can recognize a change in the space with high accuracy even if fluctuation occurs in a measurement signal obtained by measuring a physical quantity in a predetermined space. It is for the purpose.

In order to achieve the above object, a space recognition device according to the present invention is arranged on a predetermined measurement surface, measures a physical quantity of a predetermined space, and outputs a measurement signal;
A feature amount distribution generation unit that calculates a time frequency for each feature amount from each of the measurement signals in different time intervals in a small region set on the measurement surface and generates a feature amount distribution for each time interval ; ,
A change index value distribution that obtains a change index value representing a difference or similarity between the feature quantities in the small region between the feature quantity distributions having different time intervals, and uses a time frequency for each change index value as an element A change index value distribution generation unit for generating each small region ,
A spatial change recognition unit that recognizes a change in the space using the change index value distribution in the small region;
It is provided with.

Further, in the space recognition device according to the present invention, the change index value distribution generation unit includes the time frequency combination of the feature quantities between the feature quantity distributions having different time intervals for each change index value. It is also possible to calculate the time frequency of the change index value by summing up the products of the time frequencies for each combination for which the change index value is calculated.

  Furthermore, in the space recognition device according to the present invention, the space change recognition unit converts the change index value distribution of the small region using a conversion function that converts the change index value into an evaluation value representing a degree of change. A change in the space may be recognized using the evaluation value distribution in the small region by converting the time frequency for each evaluation value corresponding to each change index value into an evaluation value distribution.

Further, in the space recognition device according to the present invention, the feature amount distribution generation unit includes:
A short-time feature value distribution generation unit that calculates the time frequency for each feature value from the measurement signal in a first time interval and generates a short-time feature value distribution;
A long-time feature amount distribution generation unit that calculates the time frequency for each feature amount from the measurement signal in a second time interval longer than the first time interval and generates a long-time feature amount distribution;
With
The change index value distribution generation unit may generate the change index value distribution in the small region between the short time feature value distribution and the long time feature value distribution.

  Furthermore, in the space recognition apparatus according to the present invention, the measurement unit is arranged on the imaging surface with the measurement surface as an imaging surface, and the photoelectric conversion unit that images the space and outputs a luminance signal as the measurement signal It is good also as composition which consists of.

  According to the space recognition apparatus of the present invention, a significant difference occurs between a fluctuation component with a low time frequency and a non-fluctuation component with a high time frequency in the feature amount distribution and the change index value distribution. Therefore, it is possible to correctly recognize changes due to non-fluctuating components, and the recognition accuracy is greatly improved.

It is a figure which shows schematic structure of the space recognition apparatus which concerns on this invention. It is a schematic diagram of the distribution image obtained by the space recognition apparatus according to the present invention. It is a functional block diagram of the space recognition device concerning the present invention. (A) It is a figure which shows an example of the relationship between time and a luminance value. (B) It is a figure which shows an example of the relationship between a luminance value and time frequency. It is explanatory drawing of a short time feature-value distribution and a long-time feature-value distribution. (A)-(e) It is a figure which shows the short time feature-value distribution, long-time feature-value distribution, change index value distribution, index value / evaluation value conversion function, and evaluation value distribution when there is a change in the monitoring space. . (A)-(e) It is a figure which shows the short time feature-value distribution, long-time feature-value distribution, change index value distribution, index value / evaluation value conversion function, and evaluation value distribution when there is no change in the monitoring space. (A), (b) It is a figure which shows an example of the mode value R of the change area | region likelihood in the captured image and the captured image in which the person was imaged. It is a flowchart which shows operation | movement of the imaging part and process part in the space recognition apparatus which concerns on this invention.

  Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to FIGS. 1 to 9 of the accompanying drawings.

[Configuration of space recognition device]
As shown in FIG. 1, the space recognition device 1 according to the present embodiment includes an imaging unit 2, a storage unit 3, a processing unit 4, and an output unit 5, and continuously measures a physical quantity (for example, luminance) in a predetermined space. The change in the space is recognized from the measured signal (for example, luminance signal).

  The imaging unit 2 captures a predetermined space (monitoring space) at high speed, generates a distribution image that is a collection of feature amount distributions as shown in FIG. 2 at a predetermined cycle, and outputs the distribution image to the processing unit 4. Note that each rectangular portion in the x direction (horizontal direction) and the y direction (vertical direction) in FIG. 2 corresponds to a pixel.

  The imaging unit 2 includes a plurality of photoelectric conversion units 20 (20-1, 20-2,..., 20-N) illustrated in FIG. Each of the plurality of photoelectric conversion units 20-1, 20-2,..., 20 -N is a measurement unit, and photoelectrically converts reflected light from an object existing in the monitoring space, according to the brightness of the reflected light. Outputs a luminance signal.

  The imaging unit 2 includes a feature amount distribution generation unit 21 illustrated in FIG. The feature quantity distribution generation unit 21 holds the luminance signals from the plurality of photoelectric conversion units 20-1, 20-2,... The time frequency for each feature quantity with the photoelectric conversion unit 20 as a unit is calculated as a feature quantity distribution.

  In this example, a plurality of feature amount distributions with one or a plurality of photoelectric conversion units 20 as a unit are calculated, and an image obtained by the collection is used as a distribution image.

  The storage unit 3 includes a memory device such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and stores various programs and various data. The storage unit 3 is connected to the processing unit 4 and inputs / outputs such information to / from the processing unit 4.

  The processing unit 4 is configured by an arithmetic device such as a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or an MCU (Micro Control Unit).

  The processing unit 4 is connected to the storage unit 3 and operates as various processing / control units by reading out and executing a program from the storage unit 3. The processing unit 4 stores various data in the storage unit 3 and reads out the various data.

  Further, the processing unit 4 is also connected to the imaging unit 2 and the output unit 5, processes the distribution image from the imaging unit 2 to recognize the change area in the monitoring space, and whether the recognized change area is caused by an intruder. If it is determined that the area has been changed by an intruder, an alarm signal is output to the output unit 5.

  The output unit 5 is connected to the processing unit 4 and outputs an alarm signal from the processing unit 4 to the outside. Specifically, the output unit 5 includes a communication device and transmits an alarm signal to a remote security center via a predetermined network.

  Further, the configuration of each unit will be described with reference to the functional block diagram of FIG.

  Each of the plurality of photoelectric conversion units 20-1, 20-2,..., 20-N includes an image sensor, an electronic shutter, an amplifier, and an A / D converter. The electronic shutter, amplifier, and A / D converter may be shared by two or more photoelectric conversion units 20.

  The imaging device is composed of a CCD (Charge Coupled Devices) or a C-MOS (Complementary Metal Oxide Semiconductor), and measures the intensity of visible light in the monitoring space and outputs a luminance signal having the measured intensity as a voltage value. .

  For example, each photoelectric conversion unit 20 operates an electronic shutter every 1/500 seconds, amplifies the voltage of the imaging device at the time of operation with an amplifier, and samples with an A / D converter, thereby 1/500 seconds. A digital luminance signal in which the luminance value is discretized into 256 gradations from 0 to 255 is output every time.

  The plurality of image sensors are arranged on a predetermined imaging surface to form a surface, and project the monitoring space onto the surface. For example, the image sensors are arranged in an array on a flat substrate such that 640 elements are equally spaced in the X direction of the substrate and 480 elements are equally spaced in the Y direction, and 640 × 480 pixels every 1/500 second. The gray image is taken.

  In addition, a plurality of image pickup devices each include a set of image pickup devices each having three image pickup devices each having a color filter suitable for each of the R, G, and B bands on a flat substrate in the X direction. 640 sets at intervals and 480 sets at equal intervals in the Y direction can be arranged in an array to pick up a color image of 640 × 480 pixels.

  Further, RGB luminance signals may be converted into another color system such as HSV and output.

  The imaging cycle (measurement cycle) of the photoelectric conversion unit 20 only needs to be sufficiently shorter than a change to be detected. For example, when a change by a person is a main detection target, it may be 1/100 second or 1/200 second. Good.

  The number of gradations and the number of photoelectric conversion units 20 are not limited to the above example, and can be set according to the detection target and the size of the space.

  As described above, the plurality of photoelectric conversion units 20-1, 20-2,..., 20-N are arranged on a predetermined imaging surface, image a predetermined space, and each output a luminance signal. That is, the plurality of measurement units are arranged on a predetermined measurement surface, measure physical quantities in a predetermined space, and each output a measurement signal.

  In the present embodiment, the imaging unit 2 functions including the feature amount distribution generation unit 21, and the processing unit 4 functions including the change index value distribution generation unit 22, the space change recognition unit 24, and the intrusion detection unit 25. The storage unit 3 functions including the index value / evaluation value conversion function storage unit 23.

  The feature amount distribution generation unit 21 includes a buffer and an arithmetic circuit. The buffer is composed of, for example, a ring buffer, holds the luminance signals from the plurality of photoelectric conversion units 20-1, 20-2,..., 20-N for a certain period, and circulates and stores the luminance values for each imaging cycle.

  In this example, the number of buffers equal to or greater than a predetermined frequency calculation time width is provided as many as the number of photoelectric conversion units 20.

  For example, if the imaging cycle is 1/500 seconds and the second time width described later is 1 minute, buffers having a buffer length of 30000 or more are prepared for the number of photoelectric conversion units 20.

  The arithmetic circuit calculates the time frequency for each feature amount from the luminance signal of the frequency calculation time width held in the buffer in the frequency calculation cycle, generates a feature amount distribution, and outputs the feature amount distribution to the change index value distribution generation unit 22.

  For example, the arithmetic circuit calculates a normalized histogram of feature amounts in each of a plurality of small regions set in advance on the imaging surface of the imaging unit 2.

  For example, the small area can be the pixel of the gray image, and the feature amount can be the luminance value of the gray image. That is, the photoelectric conversion unit 20 and the small area have a one-to-one correspondence, and the feature amount is a scalar.

  Further, for example, the small area may be the pixel of the color image, and the feature amount may be the R, G, and B luminance values. In this case, the photoelectric conversion unit 20 and the small region have a one-to-one correspondence, and the feature amount is a three-dimensional vector.

  Furthermore, for example, a plurality of neighboring pixels (2 × 2 pixels, 3 × 3 pixels, etc.) in the grayscale image or color image may be set as a small region. In this case, the original data for calculating the time frequency in each small area can be increased while performing resolution conversion.

  The frequency calculation cycle is 1/5 second with respect to a first time width 1/5 second described later, that is, there is no overlap.

  Note that the frequency calculation cycle may be set shorter than the first time width and an overlap may be provided, or the cycle may be set longer than the first time width and thinned.

  As described above, the feature amount distribution generation unit 21 sequentially starts from a measurement signal (luminance signal) having a predetermined time width in each of a plurality of small regions set on the measurement surface (on the image pickup surface of the image pickup unit 2). A time frequency for each feature amount is calculated to generate a feature amount distribution.

  Here, FIG. 4A shows an example of the relationship between the luminance signal time and the luminance value in a certain pixel. As shown in FIG. 4A, it is assumed that the luminance value changes with time in the range of 0-255. When the number of luminance values of the frequency calculation time width T is 100, for example, the frequency value of each luminance value is counted, and when the time frequency is calculated by dividing each of the counting results by 100, the luminance value and time in the pixel are calculated. The frequency has a relationship like a normalized histogram shown in FIG.

  As shown in FIG. 3, the feature amount distribution generation unit 21 includes a short-time feature amount distribution generation unit 210 and a long-time feature amount distribution generation unit 211, and generates two types of feature amount distributions having different time widths.

  The short time feature amount distribution generation unit 210 calculates a time frequency for each feature amount from the luminance signal having the first time width Ta shown in FIG. 5 in each of the plurality of small regions, and generates a short time feature amount distribution. It outputs to the change index value distribution generation unit 22.

  For example, by reading out the latest 1/5 second luminance signal (time series luminance value) from the buffer for each pixel every 1/5 second, and calculating the time frequency for each luminance value from the read luminance signal, A short-time feature distribution is generated.

  The long-time feature value distribution generation unit 211 calculates the time frequency for each feature value from the luminance signal having the second time width Tb longer than the first time width Ta shown in FIG. A temporal feature quantity distribution is generated and output to the change index value distribution generation unit 22.

  For example, by reading out the latest 1 minute luminance signal (time series luminance value) from the buffer every 1/5 second and calculating the time frequency for each luminance value from the read luminance signal, Generate a feature distribution.

  The time interval for generating the short-time feature value distribution may be included in the time interval for generating the long-time feature value distribution.

  In each feature quantity distribution, the time frequency of the true element (true feature quantity) that is a non-fluctuating component shows a significantly high value, and compared with this, a false element (fake feature quantity) that is a fluctuation component ) Time frequency is sufficiently low.

  For example, even if sudden noise occurs and its amplitude is large (that is, even if it is a value greatly deviating from the true luminance value), the time frequency in the feature quantity distribution is sufficiently low, and true It is possible to distinguish the luminance value and noise.

  In the short-time feature quantity distribution, the time frequency of the true feature quantity as a non-fluctuating component is obtained for background objects such as stationary objects such as buildings, vibration / rotation objects such as plants, and moving objects such as people and vehicles. It shows a significantly high value, and the time frequency of the fake feature amount that is a fluctuation component is sufficiently low.

  On the other hand, in the long-time feature amount distribution, the time frequency of the true feature amount that is a non-fluctuation component is significantly high for the background object, and the time frequency of the false feature amount that is a moving object and the fluctuation component Is sufficiently low. In particular, due to the long-time feature distribution, not only a stationary object but also a vibration / rotation object can be treated as a non-fluctuating component of a background object.

  The change index value distribution generation unit 22 obtains a change index value representing a difference or similarity between feature quantities in a small area between the feature quantity distributions sequentially generated by the feature quantity distribution generation unit 21, and determines the change index value for each change index value. A change index value distribution having time frequency as an element is calculated.

  The change index value is a value serving as an index for recognizing the change area. For example, the feature value can be a luminance value of a grayscale image, and the change index value can be a luminance difference. In this case, the luminance difference that is the change index value is a value that represents the difference between the feature amounts. Accordingly, the difference between the small areas having a large luminance difference is high, and the probability of being a change area is high. The difference between the small areas having a luminance difference of 0 or near 0 is low, and the accuracy of being a change area is low. It will be.

  The change index value distribution is information that serves as a basis for recognizing a change area in consideration of time frequency. Specifically, for each change index value, the product of the time frequency for each combination for which the change index value is calculated among the combinations of the time frequency of the feature quantities between the feature quantity distributions is summed, and the change index value Calculate the time frequency.

  For example, when the feature value is the luminance value of the grayscale image and the change index value is the luminance difference, for each luminance difference, for each combination for which the luminance difference is calculated among the luminance frequency temporal combinations between the luminance value distributions. The product of the time frequency is summed to calculate the time frequency of the brightness difference, and the brightness difference distribution is obtained.

  Further, the short-time feature value distribution generated by the short-time feature value distribution generation unit 210 and the long-time feature value distribution generated by the long-time feature value distribution generation unit 211 so that the vibration / rotation object is not mistakenly recognized as a change region. It is preferable to calculate the change index value distribution for each small region between the two.

  For example, the time frequency Pa (i) of the luminance value i at an arbitrary pixel (x, y) constituting the short-time luminance value distribution and the time of the luminance value j at the corresponding pixel (x, y) of the long-time luminance value distribution The luminance difference distribution Ps (| i−j |) with the frequency Pb (j) is expressed by the following formula (1).

  Ps (| i−j |) = ∫j · ∫i · Pb (j) · Pa (i) · di · dj Expression (1)

  The value ranges of the luminance values i, j and the luminance difference | i−j | between i and j are all 0 to 255. Further, ∫j∫ididj means summing up the products of time frequencies for the combination of luminance values i and j whose luminance difference is | i−j |.

Here, the element Ps (0) having a luminance difference of 0 in the luminance difference distribution Ps will be described.
The luminance value i on the short-time luminance value distribution side where the luminance difference | i−j | is 0 for the luminance value j = 0, 1,..., 255 on the long-time luminance value distribution side is 0, 1,. , 255, the product of the time frequency Pb (0) on the long-time luminance value distribution side and the time frequency Pa (0) on the short-time luminance value distribution side, and the time frequency Pb (1) on the long-time luminance value distribution side, The product of the time frequency Pa (1) on the short-time luminance value distribution side, ..., the product of the time frequency Pb (255) on the long-time luminance value distribution side and the time frequency Pa (255) on the short-time luminance value distribution side Thus, the element Ps (0) is obtained.

Next, the element Ps (1) having a luminance difference of 1 in the luminance difference distribution Ps will be described.
The luminance value i on the short-time luminance value distribution side where the luminance difference | i−j | is 1 for the luminance value j = 0, 1, 2,. , 2, 1 and 3, ..., 254, the product of Pb (0) and Pa (1), the product of Pb (1) and Pa (0), the product of Pb (1) and Pa (2), The product of Pb (2) and Pa (1), the product of Pb (2) and Pa (3),..., The product of Pb (255) and Pa (254) is summed to obtain the element Ps (1). The same applies to Ps (2) to Ps (255).

  Here, FIG. 6 shows a short-time feature value distribution, a long-time feature value distribution, a change index value distribution, an index value / evaluation value conversion function, an evaluation value when there is a change in the monitoring space, such as when a person appears. An example of distribution is shown. FIG. 7 shows an example of the short-time feature value distribution, long-time feature value distribution, change index value distribution, index value / evaluation value conversion function, and evaluation value distribution when there is no change in the monitoring space. FIG. 7 shows that the luminance signal fluctuates even if there is no change in the monitoring space. Further, comparing FIG. 6 and FIG. 7, when there is a change in the monitoring space, the brightness value indicating the peak time frequency is different in the short-time feature distribution and the long-time feature distribution. When there is no change, the luminance values indicating the peak time frequency in the short-time feature amount distribution and the long-time feature amount distribution substantially coincide with each other.

  In the change index value distribution, if there is a true change between small areas, the time frequency of elements that are truly different or elements that are truly similar shows a significantly high value. The time frequency of an element showing a false difference or false similarity mixed with components is sufficiently low.

  The index value / evaluation value conversion function storage unit 23 stores in advance an index value / evaluation value conversion function for converting the change index value into an evaluation value representing the likelihood of change.

  For example, the evaluation value R represents the change area like small area, and the value range is 0 to 1. Further, a change index value having a higher change area likelihood is associated with a higher evaluation value R. Further, the evaluation value R = 0 if there is no change area, the evaluation value R = 1 if the change area is sure, and 0 <R <1 if there is ambiguity.

  As an example of a continuous function for realizing this, considering that there is ambiguity due to fluctuation in the luminance difference k near 0, if k = 0, the evaluation value R = 0, and if 0 <k <Tk, R Is a simple increase function of 0 <R <1 for k, and R = 1 if k ≧ Tk.

  Note that Tk is set according to the assumed fluctuation magnitude. For example, based on a prior experiment, Tk = 40 or the like is determined for the luminance difference k whose value range is 0 to 255. Alternatively, as an example of a step function for realizing this, if the luminance difference k is 0, the evaluation value R = 0, and if the luminance difference k is greater than 0, R = 1.

  That is, all of the above examples are conversion functions that assign a higher evaluation value that represents the change area as the luminance difference increases.

  The space change recognition unit 24 recognizes a change in the monitoring space based on the change index value distribution in each small area. For example, the space change recognition unit 24 recognizes a change region in the monitoring space.

  Specifically, the space change recognition unit 24 first reads the conversion function stored in the index value / evaluation value conversion function storage unit 23, and calculates the time frequency of the change index value according to the following equation (2). The change index value distribution of each small area is converted into an evaluation value distribution by converting the time frequency of the evaluation value corresponding to the value.

  P (R) = ∫k · Ps (R (k)) · dk (2)

  Here, k is the change index value, and Ps is the time frequency of the change index value. Further, ∫k dk means the sum total regarding the change index value k.

  The change index value is converted into an evaluation value by the conversion function R (k), and the time frequencies of the change index values having the same assigned evaluation value are integrated and converted into the time frequency of the evaluation value.

For example, k is a luminance difference, and k = | i−j |.
That is, in the example in which the conversion function is the continuous function, P (0) = Ps (R (0)),..., P (0.05) = Ps (R (1)),. Ps (R (Tk)) + Ps (R (Tk + 1)) +... + Ps (R (255)).

  Next, the space change recognition unit 24 recognizes, as a change area, a small area in which an evaluation value distribution with a mode value (an evaluation value for which the highest time frequency is calculated) is greater than 0 is calculated.

  Further, when the conversion function is the continuous function, the luminance difference k is in the vicinity of 0 and the luminance difference including the ambiguity due to the fluctuation is dispersed in the region where the evaluation value 0 <R <1, and therefore, R = 1 is inappropriately set. Can be prevented from being accumulated. Therefore, it is possible to separate the change index value based on the non-fluctuation component from the change index value based on the fluctuation component with high accuracy and recognize the change region with high accuracy.

  Alternatively, when the conversion function is the above step function, a small region in which a change index value distribution having a higher time frequency when the luminance difference is 0 is higher than a time frequency when the luminance difference is not 0 is recognized as the changed region. Will be. Even in this case, the change index value based on the non-fluctuation component can be accurately separated from the change index value based on the fluctuation component, and the change region can be recognized with high accuracy.

  Here, FIGS. 8A and 8B show an example of a distribution image in which a person is imaged and the mode value Rf of the evaluation value R in the distribution image. In the AA line of the captured image shown in FIG. 8A, as shown in FIG. 8B, the mode value Rf at the position corresponding to the human torso and both arms shows 1 and others. The mode value Rf at the position of 0 indicates 0.

  The space change recognition unit 24 may recognize the change by directly analyzing the change index value distribution without using the conversion function. In this case, for example, the spatial change recognition unit 24 recognizes the small region as a change region if the mode value in the change index value distribution in the small region is non-zero, and recognizes that the small region is not a change region if it is 0. Can do. Even in this case, the change index value based on the non-fluctuation component can be accurately separated from the change index value based on the fluctuation component, and the change region can be recognized with high accuracy.

  In addition, it is good also as a change area | region which put these together by connecting adjacent small areas among the small areas recognized as a change area. For example, among the pixels recognized as the change region, the same label is sequentially given to the pixels in the vicinity of eight, and the pixel group for each label is used as the change region information.

  The space change recognition unit 24 outputs the coordinate information of each change region recognized as described above to the intrusion detection unit 25.

  When the size of the change area is in a predetermined range, the intrusion detection unit 25 generates a predetermined alarm signal by detecting the presence of an intruder in the monitoring space, and sends the generated alarm signal to the output unit 5. Output.

  The space change recognition unit 24 may add and output the short-time feature value distribution, long-time feature value distribution, and evaluation value of each change region.

  In that case, in addition to the size of the change area, the intrusion detection unit 25 compares the characteristics of the short-time feature quantity distribution in the change area and the comparison results of the features of the short-time feature quantity distribution and the long-time feature quantity distribution in the change area, The shape of the change area or the movement of the change area may be used as the detection condition.

  In that case, the intrusion detection unit 25 obtains a difference from the reference value for the size, feature comparison result, shape, and movement, and an intruder exists when the difference weighted by the evaluation value exceeds the threshold value. Then it can be determined.

  Next, operations of the imaging unit 2 and the processing unit 4 in the space recognition device 1 configured as described above will be described with reference to the flowchart of FIG.

  Each photoelectric conversion unit 20 (20-1, 20-2,..., 20-N) of the image pickup unit 2 performs A / D conversion on the voltage of the image pickup device every time the electronic shutter is operated to obtain a luminance value. Sampling is performed, and the luminance value sampled is written into the buffer by the feature amount distribution generation unit 21 of the imaging unit 2 to buffer the luminance signal (S1).

  The feature amount distribution generation unit 21 determines whether or not the frequency calculation cycle has arrived based on the number of times of writing to the buffer or the latest writing position in the buffer (S2).

  For example, if the imaging cycle is 1/500 seconds and the frequency calculation cycle is 1/5 second, the frequency calculation cycle comes every time 100 luminance values are written.

  If the frequency calculation cycle has not arrived (NO in S2), the feature amount distribution generation unit 21 returns the process to step S1 and continues buffering.

  On the other hand, when the frequency calculation cycle arrives (YES in S2), the feature amount distribution generation unit 21 advances the process to step S3.

  The feature amount distribution generation unit 21 operates as the short time feature amount distribution generation unit 210, reads a luminance signal having a first time width that ends at the latest writing position from the buffer for each small region, and sets a luminance value for each small region. Is calculated to generate a short-time feature amount distribution (S3).

  For example, if the imaging cycle is 1/500 seconds, the first time width is 1/5 seconds, the luminance value is an integer from 0 to 255, and the small area is a pixel, the luminance value is 0 from 100 luminance values for each pixel. , 1,..., 255 are counted, and a normalized histogram obtained by dividing each count result by 100 is calculated as a short-time luminance value distribution.

  Next, the feature amount distribution generation unit 21 operates as the long-time feature amount distribution generation unit 211, reads out a luminance signal having a second time width that ends at the latest writing position from the buffer for each small region, The luminance frequency time frequency is calculated to generate a long-time feature amount distribution (S4).

  For example, if the imaging cycle is 1/500 seconds, the second time width is 1 minute, the luminance value is an integer of 0 to 255, and the small area is a pixel, the luminance value is 0,1 from 30000 luminance values for each pixel. ,..., 255 are counted, and a normalized histogram obtained by dividing each count result by 30000 is calculated as a long-time luminance value distribution.

  Alternatively, instead of buffering a long-time luminance signal, the short-time feature distribution can be buffered, and the short-time feature distribution can be integrated and normalized for each small region to generate a long-time feature distribution. it can.

  The imaging unit 2 outputs the generated short-time feature value distribution and long-time feature value distribution to the processing unit 4.

  In the flowchart of FIG. 9, the processes of the imaging unit 2 and the processing unit 4 are expressed as a series of processes. However, the imaging unit 2 that has output the feature amount distribution waits for the processing unit 4 to finish the processes in steps S3 to S9. Instead, the process may be returned to step S1.

  Subsequently, the change index value distribution generation unit 22 of the processing unit 4 calculates the change index value distribution for each of the input short-time feature quantity distribution and corresponding small regions of the long-time feature quantity distribution, and generates an index value distribution image. Generate (S5).

  For example, the change index value distribution generation unit 22 calculates the luminance difference distribution by applying Equation (1) for each corresponding pixel of the short-time luminance value distribution and the long-time luminance value distribution.

  Subsequently, the spatial change recognition unit 24 of the processing unit 4 reads the conversion function from the index value / evaluation value conversion function storage unit 23, applies the conversion function to the change index value distribution of each small region, and evaluates the small region. A value distribution is calculated (S6). For example, the spatial change recognition unit 24 calculates the evaluation value distribution of the pixel by applying Equation (2) to the luminance difference distribution of each pixel.

  When the mode value of the evaluation value distribution is 0 in all the small regions (the evaluation value with the highest time frequency is 0), the spatial change recognition unit 24 assumes that no change region has been detected (NO in step S7). The process returns to step S1. The processing unit 4 waits for input of the feature amount distribution.

  On the other hand, if there is a small region where the mode value of the evaluation value distribution is greater than 0, the space change recognition unit 24 generates information on the changed region, assuming that the changed region is detected (YES in step S7). Then, the process proceeds to step S8.

  For example, the spatial change recognizing unit 24 puts together pixels whose coordinates are adjacent to each other among the pixels having the mode value of the evaluation value distribution greater than 0 as one change area. The feature amount distribution and the change area likelihood are output to the intrusion detection unit 25.

  Subsequently, the intrusion detection unit 25 of the processing unit 4 compares the information on the change area generated by the space change recognition unit 24 with the intrusion detection condition (S8).

  If a change area that satisfies the intrusion detection condition is generated (YES in S8), a predetermined alarm signal is generated and input to output unit 5 (S9).

  When the alarm signal is input from the intrusion detection unit 25, the output unit 5 transmits the alarm signal to the security center.

  If a change area that satisfies the intrusion detection condition is not generated (NO in S8), the process of step S9 is skipped.

  When the above processing is completed, the processing returns to step S1, the imaging unit 2 generates the next distribution image, and the processing unit 4 waits for the input of the distribution image from the imaging unit 2.

[Modification]
In the above-described embodiment, an example in which a luminance difference, that is, a difference between feature amounts is used as a change index value representing a difference between feature amounts has been described. In another embodiment, a distance value between feature quantities can be used as a change index value representing a difference between feature quantities. Also in this case, the change index value distribution generation unit 22 sums, for each distance value, the product of the time frequencies for each combination for which the distance value is calculated among the combinations of the time frequencies of the feature amounts between the feature amount distributions. Then, by calculating the time frequency of the distance value, a distance value distribution having the time frequency for each distance value as an element is obtained. Further, the index value / evaluation value conversion function storage unit 23 stores a conversion function that assigns a higher evaluation value representing the degree of change to a larger change index value. Then, the space change recognition unit 24 uses the conversion function to convert the time frequency of the distance value into the time frequency of the evaluation value corresponding to the distance value, thereby converting the distance value distribution of each small region into the time frequency of the evaluation value. Is converted into an evaluation value distribution having elements as elements, and changes in the space are recognized based on the mode value of the evaluation value distribution in each small region.

  In another embodiment, a change index value representing similarity between feature quantities can be used. In this case, the change index value can be a similarity or a correlation value. For example, a value obtained by subtracting the above-described luminance difference from 255 is an example of similarity. Also in this case, the change index value distribution generation unit 22 calculates, for each similarity (or correlation value), the similarity (or correlation value) in which the similarity (or correlation value) is calculated among the time frequency combinations of the feature amounts between the feature amount distributions. Similarity distribution (or correlation value distribution) with the time frequency of each similarity (or correlation value) as an element by calculating the time frequency of the similarity (or correlation value) by summing up the products of time frequency for each ) In this case, the index value / evaluation value conversion function storage unit 23 stores a conversion function that assigns a higher evaluation value that represents the degree of change as the change index value becomes smaller. Then, the space change recognition unit 24 uses the conversion function to convert the time frequency of the similarity (or correlation value) into the time frequency of the evaluation value corresponding to the similarity (or correlation value), thereby converting each small region. The similarity distribution (or correlation value distribution) is converted into an evaluation value distribution using the time frequency of the evaluation value as an element, and a change in the space is recognized based on the mode value of the evaluation value distribution in each small region.

  In the above-described embodiment and each modification, the feature amount may be a luminance value, a luminance value histogram, a motion vector, or the like. When the feature quantity is a motion vector, the change index value distribution generation unit 22 calculates a motion vector that is a change index value in a range including a small area nearby between distribution images.

  Further, in the above-described embodiment and each modification, the change index value distribution generation unit 22 in FIG. 3 does not include the long-time feature amount distribution generation unit 211, and is changed by the short-time feature amount distribution generation unit 210. The change index value distribution may be calculated for each small region between the short-time feature amount distributions generated in this way.

  In the above-described embodiment and each modification, the feature amount distribution generation unit 21 may be realized in the processing unit 4. In that case, the imaging unit 2 outputs a captured image in which luminance values are arranged for each imaging cycle to the processing unit 4. The processing unit 4 operates as the feature amount distribution generation unit 21, buffers the captured image from the imaging unit 2 in the storage unit 3, and generates a feature amount distribution for each frequency calculation period.

  In the above-described embodiment and each modified example, the space change recognition unit 24 recognizes the change region, but is not limited to this, the brightness change in the space, the movement of the object (speed or / and moving direction), It can be used for recognition of various changes such as increase / decrease in the number of objects.

  In the above-described embodiment and each modification, the measurement unit outputs a luminance signal having an amplitude corresponding to the intensity of visible light in the monitoring space, but is not limited thereto.

  For example, a configuration in which the measurement unit outputs a luminance signal having an amplitude corresponding to the intensity of near infrared light in the monitoring space, a configuration in which the measurement unit outputs a luminance signal having an amplitude according to the intensity of infrared light in the monitoring space, and the measurement unit A configuration for outputting a luminance signal having an amplitude corresponding to the intensity of ultraviolet light in the monitoring space, a configuration for the measurement unit to output a luminance signal having an amplitude corresponding to the intensity of the X-ray in the monitoring space, and a distance sensor as the measurement unit A configuration in which the measurement unit measures a predetermined physical quantity in the space and outputs a measurement signal, such as a configuration that outputs a distance signal having an amplitude corresponding to the distance from the measurement unit to an object existing in the monitoring space, may be used. As the distance sensor, for example, a method of projecting and receiving laser light, a method of transmitting and receiving electromagnetic waves such as microwaves and millimeter waves, and a method of transmitting and receiving ultrasonic waves are conceivable.

  In the above-described embodiment and each modification, an example in which a plurality of small areas are set on the entire measurement surface has been described. However, a plurality of small areas may be set on a part of the measurement surface. One small area may be set for a part.

  The best mode of the space recognition device according to the present invention has been described above, but the present invention is not limited by the description and drawings according to this mode. That is, it is a matter of course that all other forms, examples, operation techniques, and the like made by those skilled in the art based on this form are included in the scope of the present invention.

DESCRIPTION OF SYMBOLS 1 Spatial recognition apparatus 2 Imaging part 3 Memory | storage part 4 Output part 20 (20-1, 20-2, ..., 20-N) Photoelectric conversion part 21 Feature-value distribution generation part 22 Change index value distribution generation part 23 Index value / evaluation Value conversion function storage unit 24 Spatial change recognition unit 25 Intrusion detection unit 210 Short time feature amount distribution generation unit 211 Long time feature amount distribution generation unit

Claims (5)

A measurement unit arranged on a predetermined measurement surface and measuring a physical quantity of a predetermined space and outputting a measurement signal;
A feature amount distribution generation unit that calculates a time frequency for each feature amount from each of the measurement signals in different time intervals in a small region set on the measurement surface and generates a feature amount distribution for each time interval ; ,
A change index value distribution that obtains a change index value representing a difference or similarity between the feature quantities in the small region between the feature quantity distributions having different time intervals, and uses a time frequency for each change index value as an element A change index value distribution generation unit for generating each small region ,
A spatial change recognition unit that recognizes a change in the space using the change index value distribution in the small region;
A space recognition device comprising: The change index value distribution generation unit is configured to calculate, for each change index value, a time for each combination in which the change index value is calculated among the time frequency combinations of the feature quantities between the feature quantity distributions having different time intervals. The space recognition apparatus according to claim 1, wherein the frequency products are summed to calculate the time frequency of the change index value. The spatial change recognition unit uses the conversion function for converting the change index value into an evaluation value representing a degree of change, and the evaluation value corresponding to each change index value in the change index value distribution of the small region The space recognition device according to claim 1, wherein a change in the space is recognized using the evaluation value distribution in the small area by converting the evaluation value distribution having each time frequency as an element. The feature quantity distribution generation unit
A short-time feature value distribution generation unit that calculates the time frequency for each feature value from the measurement signal in a first time interval and generates a short-time feature value distribution;
A long-time feature amount distribution generation unit that calculates the time frequency for each feature amount from the measurement signal in a second time interval longer than the first time interval and generates a long-time feature amount distribution;
With
The said change index value distribution production | generation part produces | generates the said change index value distribution in the said small area | region between the said short time feature-value distribution and the said long-time feature-value distribution. Spatial recognition device. The said measurement part is arrange | positioned on this imaging surface by making the said measurement surface into an imaging surface, and consists of a photoelectric conversion part which images the said space and outputs the luminance signal as said measurement signal. The space recognition device according to any one of 4.

Images