JP5911634B1 - Image processing device - Google Patents

Abstract

An object in an image is detected easily and at high speed. An image processing apparatus according to one aspect includes a first pixel determination unit that determines a first pixel from a plurality of pixels in a processing target image, and a first pixel that is spaced apart from the first pixel in the processing target image. A second pixel determining unit that determines a plurality of second pixels arranged around the pixel, a pixel value of the first pixel determined by the first pixel determining unit, and a plurality of second pixels determined by the second pixel determining unit A pixel value acquisition unit that acquires each pixel value of the two pixels, a plurality of second pixel values based on the pixel value of the first pixel acquired by the pixel value acquisition unit and each pixel value of the plurality of second pixels. A detection unit that detects an object that does not overlap two pixels and overlaps the first pixel. [Selection] Figure 1

Description

  The present invention relates to an image processing apparatus.

  Conventionally, accidents called bird strikes have frequently occurred in which a bird collides with an artificial structure such as a windmill. In recent years, illegal flying objects such as drones have been flying over restricted areas. In order to prevent such a situation in advance or to detect it quickly, image analysis is performed on the video of a monitoring camera that monitors a predetermined monitoring area (in the above example, in the vicinity of the windmill or above the restricted entry area). Therefore, it is effective to detect an object such as a bird or a drone. This eliminates the need for the surveillance staff to constantly monitor the video from the surveillance camera.

  As an apparatus for performing image analysis for object detection as described above, for example, intruder detection that detects an intruder into a prohibited entry area by analyzing a video of a security camera installed in the prohibited entry area An apparatus is known (see Patent Document 1 below). This device divides the video of the security camera into appropriate small areas, performs correlation calculation with the background image acquired in advance for each small area, and intruders in the video based on the temporal change of the correlation value Specify the position of.

JP 2006-252248 A

  However, in the above apparatus, it is necessary to execute processing such as correlation calculation for all small areas obtained by dividing the video of the security camera. That is, in the above apparatus, since it is necessary to execute image processing for a plurality of frames (images) of the video of the security camera, the processing tends to be complicated and the processing time tends to be long. Therefore, there is a demand for a mechanism that can easily and quickly detect an object in an image.

  An image processing apparatus according to an aspect of the present invention includes a first pixel determination unit that determines a first pixel from a plurality of pixels in a processing target image, and a first pixel that is spaced apart from the first pixel in the processing target image. A second pixel determining unit that determines a plurality of second pixels arranged in the periphery, a pixel value of the first pixel determined by the first pixel determining unit, and a plurality of second values determined by the second pixel determining unit A pixel value acquisition unit that acquires each pixel value of the pixel, a plurality of second values based on the pixel value of the first pixel acquired by the pixel value acquisition unit and the pixel value of each of the plurality of second pixels. A detection unit that detects an object that does not overlap the pixel and overlaps the first pixel.

  In such a form, it is possible to detect an object having a size that fits within the region surrounded by the plurality of second pixels from the pixel values of the first pixel in the processing target image and the plurality of second pixels arranged around the first pixel. . That is, in the above-described embodiment, the pixel value of the pixel located between the first pixel and the second pixel is not used, and the first pixel and the peripheral pixel (second pixel) of the first pixel are relatively small. An object overlapping the first pixel can be detected based on the pixel values of several pixels. Moreover, in the said form, the process which compares between several flame | frame (image) continuous in time is not required. Therefore, according to the said form, the object in an image can be detected simply and at high speed.

  In the image processing apparatus according to another aspect, the second pixel determining unit may determine a plurality of second pixels based on a size of an object that is predetermined as a detection target in the processing target image.

  According to this aspect, an object predetermined as a detection target can be detected appropriately. For example, by determining the second pixel so that the object determined as the detection target is exactly within the region surrounded by the plurality of second pixels, an object that is not a detection target larger than the object determined as the detection target Unintentional detection can be prevented.

  In the image processing apparatus according to still another aspect, the detection unit has a pixel value variation of each of the plurality of second pixels equal to or less than a predetermined reference, and a representative value of each pixel value of the plurality of second pixels. When the difference between the pixel value of the first pixel and the pixel value of the first pixel is equal to or greater than a predetermined threshold, it may be determined that an object exists.

  According to this embodiment, based on the knowledge of the present inventor described below, it is possible to easily and accurately detect an object that does not overlap the plurality of second pixels but overlaps the first pixels. When each of the plurality of second pixels corresponds to a relatively uniform brightness or shade background such as the sky, the variation in the pixel values of the plurality of second pixels tends to be relatively small. Further, when there is an object that does not overlap the second pixel and overlaps the first pixel, the pixel value of the first pixel that overlaps the object and the pixel value of the plurality of second pixels that overlap the background such as the sky A difference from a representative value (for example, an average value or a median value of pixel values of a plurality of second pixels) tends to be relatively large. Therefore, by executing the determination process based on such knowledge, it is possible to easily and accurately detect an object existing in the background of relatively uniform brightness and hue.

  The image processing apparatus according to another aspect further includes a binarization unit that binarizes the pixel value of each pixel in the processing target image, and the first pixel determination unit sequentially sets each pixel in the processing target image. Each process performed by the first pixel determination unit, the second pixel determination unit, the pixel value acquisition unit, and the detection unit is performed on each pixel determined as the first pixel by the first pixel determination unit, and binarization is performed. The pixel value of each pixel in the processing target image so that the pixel value of the first pixel in which the object is detected by the detection unit and the pixel value of the first pixel in which the object is not detected by the detection unit are different from each other May be binarized.

  According to this aspect, for each case where each pixel in the processing target image is the first pixel, each process of the second pixel determination unit, the pixel value acquisition unit, and the detection unit is executed, and the object is detected. The binarization unit binarizes so that the pixel value of the pixel and the pixel value of the pixel where no object is detected have different gradations (for example, white and black). Thereby, it is possible to obtain a binarized image in which the position of the object in the processing target image is visually easy to understand.

  According to one aspect of the present invention, an object in an image can be detected easily and at high speed.

It is a block diagram which shows the function structure of the image processing apparatus which concerns on one Embodiment. It is a block diagram which shows the hardware constitutions of the image processing apparatus which concerns on one Embodiment. It is a figure which shows the example of the center pixel and peripheral pixel which are determined in a process target image. It is a figure which shows the example in which an object is detected by the detection part. 6 is a flowchart illustrating an operation of the image processing apparatus according to the embodiment.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or equivalent elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a block diagram illustrating a functional configuration of an image processing apparatus according to an embodiment. An image processing apparatus 1 illustrated in FIG. 1 is an apparatus that detects whether or not an object to be detected exists in a processing target image by executing image processing on an image input as a processing target image.

  The image processing apparatus 1 acquires, as a processing target image, an image in the vicinity of the windmill taken by a monitoring camera or the like in order to prevent a so-called bird strike in which a bird collides with the windmill, for example. Here, the processing direction and the imaging angle of the processing target image are set so that a relatively uniform brightness or shade background (for example, the sky) is reflected so that it can be easily detected by the detection unit 14 described later. It is the image image | photographed with the surveillance camera. In the case of the above example, a bird that has approached the windmill more than a certain amount is determined as an object to be detected (hereinafter “detection target object”), and the image processing apparatus 1 determines whether or not the detection target object exists in the processing target image. Set to detect.

  When sequentially receiving each frame (image) of a moving image from the surveillance camera, the image processing apparatus 1 may execute image processing, which will be described later, using frames extracted from the frames at predetermined intervals as processing target images. . By continuously executing such processing, the image processing apparatus 1 can detect an object to be detected by processing an image continuously acquired from the monitoring camera in almost real time. When detecting the detection target object, the image processing apparatus 1 may notify the monitor that the detection target object has been detected by screen output, audio output, or the like. Further, the image processing apparatus 1 may fire a blank cannon by transmitting a control signal to an unillustrated blank cannon launching apparatus installed to intimidate and move away a bird approaching the windmill or the like, for example.

  As illustrated in FIG. 1, the image processing apparatus 1 includes a central pixel determination unit (first pixel determination unit) 11 and a peripheral pixel determination unit (first pixel) as functional elements for detecting a detection target object in a processing target image. 2 pixel determination unit) 12, pixel value acquisition unit 13, detection unit 14, and binarization unit 15.

  FIG. 2 is a block diagram illustrating an example of a hardware configuration of the image processing apparatus 1. As shown in FIG. 2, the image processing apparatus 1 includes one or more CPUs (Central Processing Units) 101, a RAM (Random Access Memory) 102 that is a main storage device, and a ROM (Read Only Memory) 103, such as a surveillance camera. The computer system includes a communication module 104 for performing data communication with an external device, and hardware such as an auxiliary storage device 105 such as a hard disk. The image processing device 1 may be physically configured as a single device, or may be configured from a plurality of devices that operate in cooperation with each other. Each functional element of the image processing apparatus 1 shown in FIG. 1 causes the communication module 104 to operate under the control of the CPU 101 by reading predetermined computer software on hardware such as the RAM 102, and the RAM 102 and the ROM 103. This is realized by reading and writing data in the auxiliary storage device 105. Alternatively, each functional element of the image processing apparatus 1 illustrated in FIG. 1 may be implemented by being programmed in an integrated circuit such as an FPGA.

  The center pixel determining unit 11 is means for determining a center pixel (first pixel) from a plurality of pixels in the processing target image. As illustrated in FIG. 3, the center pixel determination unit 11 determines the center pixel C from a plurality of pixels in, for example, a rectangular processing target image T. For example, the center pixel determining unit 11 scans all the pixels in the processing target image T in a predetermined order (for example, raster scan), and determines the scanning destination pixel as the center pixel C. A series of processes of the peripheral pixel determination unit 12, the pixel value acquisition unit 13, and the detection unit 14 are executed for the center pixel C determined in this way. When the execution of the series of processes is completed, the center pixel determination unit 11 advances the scanning of the processing target image T, and is at a position moved to the right by one pixel from the next scan destination pixel (for example, the current center pixel C). Pixel) is determined as a new central pixel C. Thereafter, a series of processes of the peripheral pixel determination unit 12, the pixel value acquisition unit 13, and the detection unit 14 are executed on the newly determined center pixel C. In this manner, image processing focusing on each of all the pixels in the processing target image T is sequentially executed. However, in the processing target image T, as a region where the detection target object needs to be detected (hereinafter referred to as “necessary region”), for example, a region within a predetermined range from the windmill is determined in advance, the center The pixel determining unit 11 may determine only the pixels included in the necessary area as the central pixel C. Moreover, a series of processes of the surrounding pixel determination unit 12, the pixel value acquisition unit 13, and the detection unit 14 may be executed in parallel for each of the plurality of central pixels C set at the same time.

  The peripheral pixel determination unit 12 is a means for determining a plurality of peripheral pixels (second pixels) P that are spaced from the central pixel C and arranged around the central pixel C in the processing target image T. Here, the separation of the central pixel C and the peripheral pixel P means that the central pixel C and the peripheral pixel P are not adjacent to each other. Specifically, as shown in FIG. 3, it means that at least eight pixels located at the upper left, upper, upper right, right, lower right, lower, lower left, and left of the central pixel C do not match the peripheral pixel P. . In the present embodiment, as an example, the peripheral pixel determining unit 12 determines, as the peripheral pixel P, an outer peripheral portion of a circular operator that falls within the rectangular area A centered on the central pixel C. The operator is an area composed of a plurality of pixels including the center pixel C in the processing target image T, and is an area set in association with the center pixel C when executing image processing focusing on the center pixel C.

  In the example of FIG. 3, a region (including the peripheral pixel P) surrounded by the peripheral pixels P arranged in an annular shape is set as a circular operator centered on the central pixel C. In this example, the circular operator is set to a size that fits in a rectangular area A of 7 × 7 pixels. However, a circular operator having a size that can fit in a rectangular area having a size other than the above (for example, 32 × 32 pixels) may be set according to the pixel size and the like, and the outer peripheral portion of the circular operator may be determined as the peripheral pixel P. Further, the shape of the operator is not limited to a circle, and may be a rectangle, for example. In this case, the peripheral pixels P may be determined in a rectangular ring shape so as to correspond to the four sides of the rectangular operator. The peripheral pixel P does not necessarily correspond to the outer peripheral portion of the operator. For example, the peripheral pixel determining unit 12 may determine a pixel located inside the outer peripheral portion of the operator as the peripheral pixel P.

  When the pixel in the vicinity of the edge of the processing target image T is set as the center pixel C, a part of the above-described rectangular area A protrudes outside the processing target image T, and as shown in the example of FIG. There may occur a case where the number of surrounding pixels P cannot be determined. Therefore, such pixels near the edge of the processing target image T may be excluded from the processing target pixels (targets determined as the central pixel C by the central pixel determination unit 11). Alternatively, when such a pixel is set as the central pixel C, the peripheral pixel determining unit 12 selects the peripheral pixel P whose relative position with respect to the central pixel C is determined by a predetermined rule as illustrated in FIG. Only the peripheral pixels P arranged in the processing target image T may be determined. In this case, the center pixel C is surrounded by the edge of the processing target image T and the plurality of peripheral pixels P, and the “region surrounded by the peripheral pixels P” in the following description is “the peripheral pixels P”. And “region surrounded by the edge of the processing target image T”.

  Here, the surrounding pixel determination unit 12 may determine a plurality of surrounding pixels P based on the size in the processing target image T of an object (detection target object) previously determined as a detection target. For example, the peripheral pixel determination unit 12 stores in advance the size of the detection target object in the processing target image T (for example, the number of pixels that the detection target object having an average size occupies in the processing target image T). As a parameter, the position of the relative peripheral pixel P relative to the central pixel C may be calculated by a predetermined calculation. Alternatively, the peripheral pixel determination unit 12 sets setting information (relative peripheral pixel P relative to the central pixel C as a reference) determined in advance by the user of the image processing apparatus 1 based on the size of the detection target object in the processing target image T. A plurality of peripheral pixels P may be determined by reading

  For example, consider a case where a bird (detection target object) that approaches a monitoring target such as a windmill more than a certain distance is detected. In this case, “the size of the detection target object in the processing target image T” means an approximate number of pixels that the processing target image T occupies when a bird approaching the windmill more than a certain distance appears in the processing target image T. The number of pixels is, for example, a pixel size (vertical width and horizontal width per pixel) preset for the processing target image T, an average size of a detection target object (in this case, a bird), and monitoring. It is set in advance by the user of the image processing apparatus 1 based on the distance between the camera and a monitoring object such as a windmill.

  For example, the surrounding pixel determination unit 12 may determine the surrounding pixels P so that the detection target object is just within the region surrounded by the plurality of surrounding pixels P. “Fits exactly within the region surrounded by the plurality of peripheral pixels P” means that when the approximate center of the detection target object overlaps the central pixel C, the detection target object does not overlap the peripheral pixel P, while the detection target object When the approximate center does not overlap the center pixel C, it means that the detection target object overlaps any one of the peripheral pixels P.

  FIG. 4 shows an example in which the detection target object O just fits in a region surrounded by a plurality of peripheral pixels P. In this example, as information indicating the size of the detection target object O in the processing target image T, for example, information indicating that the average width (the width from the left wing end to the right wing end) is 5 pixels is set in advance. . Based on this information, the peripheral pixel determining unit 12 determines the peripheral pixels P so that there are five pixels between the peripheral pixels P facing each other in the vertical direction or the horizontal direction. By determining the peripheral pixel P in this manner, an object larger than the detection target object O (for example, an object that exists near the monitoring camera and is located away from the monitoring object). It can prevent that it detects by the detection part 14 mentioned later unintentionally.

  The pixel value acquisition unit 13 obtains the pixel value Vc of the central pixel C determined by the central pixel determination unit 11 and the pixel values Vp1 to VpN of each of the plurality of peripheral pixels P determined by the peripheral pixel determination unit 12. It is a means to acquire. Here, N is the number of peripheral pixels P, and N is 16 in the example of FIG. Further, the pixel value is a numerical value (gradation value) that expresses gradation and strength such as pixel color (hue) and brightness (illuminance) in steps. The number of pixel value stages depends on the number of bits representing the pixel value. For example, when the pixel value is represented by 8 bits, the pixel value is represented by 256 levels from 0 to 255.

  Based on the pixel value Vc of the center pixel C acquired by the pixel value acquisition unit 13 and the pixel values Vp1 to VpN of each of the plurality of peripheral pixels P, the detection unit 14 does not overlap the plurality of peripheral pixels P and is centered This is means for detecting a detection target object O that overlaps the pixel C.

  When each of the plurality of peripheral pixels P corresponds to a background of relatively uniform brightness and color, such as the sky, the variation in the pixel values Vp1 to VpN of each of the plurality of peripheral pixels P tends to be relatively small. There is. Further, when there is a detection target object O that does not overlap the plurality of peripheral pixels P and overlaps the center pixel C, a plurality of peripherals that overlap the pixel value of the center pixel C that overlaps the detection target object O and the background such as the sky. The difference between the pixel values Vp1 to VpN of the pixel P and the representative value R (for example, the average value or median value of the pixel values Vp1 to VpN) tends to be relatively large. Based on the above knowledge, the detection unit 14 detects the detection target object O as follows.

First, the detection unit 14 determines whether or not the variation in the pixel values Vp1 to VpN of each of the plurality of peripheral pixels P is equal to or less than a predetermined reference (hereinafter also referred to as “variation determination”). In this embodiment, as an example, the detection unit 14 extracts the maximum value max and the minimum value min of the pixel values Vp1 to VpN, and the difference “max−min” between the maximum value max and the minimum value min is a predetermined threshold value d1 ( It is determined whether or not d1> 0) or less. Specifically, the detection unit 14 determines whether or not the following formula (1) is established.
max-min ≦ d1 (1)

  The detection unit 14 determines that the variation in the pixel values Vp1 to VpN is equal to or less than a predetermined reference when the above formula (1) is satisfied, and the pixel values Vp1 to Vp1 when the above formula (1) is not satisfied. It is determined that the variation in VpN is not less than a predetermined reference. In addition, the detection part 14 may perform dispersion | variation determination by methods other than the determination using said Formula (1). For example, the detection unit 14 may calculate the variance of the pixel values Vp1 to VpN and perform the variation determination depending on whether the variance is equal to or less than a predetermined threshold. Specifically, when the variance of the pixel values Vp1 to VpN is equal to or less than a predetermined threshold, the detection unit 14 determines that the variation in the pixel values Vp1 to VpN is equal to or less than a predetermined reference, and the pixel values Vp1 to VpN. If the variance of the pixel values Vp1 to VpN is not equal to or less than a predetermined threshold, it may be determined that the variation of the pixel values Vp1 to VpN is not equal to or less than a predetermined reference.

When it is determined by the variation determination that the variations in the pixel values Vp1 to VpN are equal to or less than a predetermined reference, the detection unit 14 determines the difference between the representative value R of the pixel values Vp1 to VpN and the pixel value Vc of the center pixel C (| Whether or not R−Vc |) is equal to or greater than a predetermined threshold value d2 (d2> 0) (hereinafter also referred to as “difference determination”) is executed. Specifically, the detection unit 14 determines whether or not the following formula (2) is satisfied. Here, the representative value R is, for example, an average value or a median value of the pixel values Vp1 to VpN of the peripheral pixels P. However, since the pixel values Vp1 to VpN of the peripheral pixel P are considered to be close to each other when the above formula (1) is established, the detection unit 14 determines, for example, from the pixel values Vp1 to VpN of the peripheral pixel P. Randomly acquired pixel values may be used as the representative value R.
| R−Vc | ≧ d2 (2)

  The detection unit 14 determines that there is a detection target object O that does not overlap the plurality of peripheral pixels P and overlaps the center pixel C when both of the expressions (1) and (2) are satisfied. That is, the detection unit 14 has a variation in the pixel values Vp1 to VpN of each of the plurality of peripheral pixels P equal to or less than a predetermined reference, and a difference (| R− between the representative value R and the pixel value Vc of the central pixel C). When Vc |) is equal to or greater than a predetermined threshold value d2, it is determined that the detection target object O exists. By executing the determination process (variation determination and difference determination), it is possible to easily and accurately detect the detection target object O present in the background of relatively uniform brightness and hue. In addition, the detection procedure by the detection part 14 is not restricted to the above-mentioned procedure. For example, the detection unit 14 may perform the variation determination after performing the difference determination. For the binarization processing of the binarization unit 15 described later, the detection unit 14 is, for example, a flag indicating that the detection target object O has been detected at the center pixel C determined that the detection target object O exists. Information may be added.

  The binarization unit 15 is configured such that the pixel value of the central pixel C where the detection target object O is detected by the detection unit 14 and the pixel value of the central pixel C where the detection target object O is not detected by the detection unit 14 are different from each other. In addition, it is a means for binarizing the pixel value of each pixel in the processing target image T. For example, the binarization unit 15 converts the pixel value Vc of the center pixel C where the detection target object O is detected into white (255 in the case of 8 bits), and the center pixel where the detection target object O is not detected The C pixel value Vc is converted to black (0). For example, the binarization unit 15 converts the pixel value of the pixel to which the flag information is added by the detection unit 14 to white, and converts the pixel value of the pixel to which the flag information is not added by the detection unit 14 to black Thus, the conversion process can be executed. Thereby, a binarized image in which the position of the detection target object O in the processing target image T is visually easy to understand can be obtained. In the case of the above processing example, in the processing target image T, it is possible to obtain a binarized image in which only the portion where the detection target object O is detected is expressed in white and the other portions are expressed in black.

  Subsequently, the operation of the image processing apparatus 1 will be described with reference to FIG.

  First, as shown in FIG. 3, the center pixel determining unit 11 determines the center pixel C from a plurality of pixels in the processing target image T (step S1). For example, the center pixel determining unit 11 scans pixels in the processing target image T in a predetermined scanning order (for example, raster scan), and determines a pixel to be scanned as the center pixel C. Subsequently, the peripheral pixel determination unit 12 determines a plurality of peripheral pixels P that are arranged around the center pixel C and are separated from the center pixel C in the processing target image T (step S2). Subsequently, the pixel value acquisition unit 13 acquires the pixel value Vc of the center pixel C and the pixel values Vp1 to VpN of each of the plurality of peripheral pixels P (step S3).

  Subsequently, the detection unit 14 determines the presence or absence of the detection target object O by executing the determination process (variation determination and difference determination) (steps S4 to S6). First, the detection unit 14 performs the above-described variation determination (step S4). As an example, the detection unit 14 determines whether or not the difference between the maximum value max and the minimum value min of the pixel values Vp1 to VpN of the peripheral pixels P is equal to or less than the threshold value d1 by the above formula (1).

  When the difference between the maximum value max and the minimum value min is equal to or less than the threshold value d1 (step S4: YES), the detection unit 14 continues to perform difference determination (step S5). As an example, the detection unit 14 determines whether or not the difference between the representative value R of the pixel values Vp1 to VpN of the peripheral pixel P and the pixel value Vc of the center pixel C is equal to or greater than the threshold value d2 by the above equation (2). To do. When the difference between the representative value R of the pixel values Vp1 to VpN of the peripheral pixel P and the pixel value Vc of the central pixel C is greater than or equal to the threshold value d2, the detection unit 14 does not overlap with the peripheral pixels P and the center It is determined that the detection target object O that overlaps the pixel C exists. That is, the detection unit 14 detects the presence of the detection target object O (step S6). At this time, the detection unit 14 adds flag information indicating that the detection target object O has been detected to the center pixel C for the binarization processing of the binarization unit 15.

  When the difference between the maximum value max and the minimum value min is not less than or equal to the threshold value d1 (step S4: NO), or the difference between the representative value R of the pixel values Vp1 to VpN of the peripheral pixel P and the pixel value Vc of the center pixel C is If it is not equal to or greater than the threshold value d2 (step S5: NO), the detection unit 14 determines that there is no detection target object O that does not overlap the plurality of peripheral pixels P and overlaps the center pixel C. That is, the detection unit 14 does not detect the presence of the detection target object O.

  The processes in steps S1 to S6 are executed when all the pixels included in the processing target image T (or the necessary area described above) are the central pixel C (step S7: NO). When a series of processes for all the pixels is completed (step S7: YES), the binarization unit 15 detects the pixel value of the central pixel C in which the detection target object O is detected by the detection unit 14 and the detection target by the detection unit 14. The pixel values of the respective pixels in the processing target image T are binarized so that the pixel values of the central pixel C where the object O is not detected are different from each other. As an example, the binarization unit 15 converts the pixel value Vc of the center pixel C where the detection target object O is detected into white (255 in the case of 8 bits), and the center where the detection target object O is not detected. The pixel value Vc of the pixel C is converted to black (0). For example, the binarization unit 15 converts the pixel value of the pixel to which the flag information is added by the detection unit 14 in Step S6 into white, and sets the pixel value of the pixel to which the flag information is not added by the detection unit 14 to black The conversion process can be executed by conversion. Thereby, a binarized image in which the position of the detection target object O in the processing target image T is visually easy to understand can be obtained.

  Note that the operation of the image processing apparatus 1 illustrated in FIG. 5 is an example, and part of the processing content may be changed, or part of the processing order may be changed. For example, the execution order of the variation determination in step S4 and the difference determination in step S5 may be switched. Further, instead of the process of step S4, the determination based on the dispersion of the pixel values Vp1 to VpN of the peripheral pixels P described above may be executed. In the above example, the binarization process is performed after a series of processes (steps S1 to S6) for all the pixels in the processing target image (or necessary area) is completed. May be performed each time a series of processing for each pixel is completed. That is, the binarization unit 15 may perform binarization processing on the central pixel C at that time, for example, before step S7 (before the next central pixel C is determined).

  According to the image processing apparatus 1 described above, within the region surrounded by the plurality of peripheral pixels P from the pixel values of the plurality of peripheral pixels P arranged at and around the center pixel C in the processing target image T. It is possible to detect the detection target object O having a size that fits. That is, the image processing apparatus 1 does not use the pixel value of the pixel located between the central pixel C and the peripheral pixel P, and the pixel value of a relatively small number of pixels including the central pixel C and the peripheral pixel P. Based on the above, it is possible to detect the detection target object O overlapping the center pixel C. Further, the image processing apparatus 1 does not require processing for comparing a plurality of temporally continuous frames (images). Therefore, according to the image processing apparatus 1, the detection target object O in the processing target image T can be detected easily and at high speed.

  More specifically, according to the image processing apparatus 1, image processing can be performed easily and at high speed as compared with a case where general image processing (that is, image processing based on pixel values of all pixels in the operator) is executed. Can be executed. Further, according to the image processing apparatus 1, it is not necessary to use temporal changes in pixel values between a plurality of temporally consecutive frames in order to detect an object in an image. Compared to the case where image processing based on temporal change of pixel values is executed, the image processing can be executed easily and at high speed.

  The present invention has been described in detail based on the embodiments. However, the present invention is not limited to the above embodiment. The present invention can be variously modified without departing from the gist thereof.

  For example, the use of the image processing apparatus 1 is not limited to the use for preventing the above-described bird strike. The image processing apparatus 1 may be used, for example, for obtaining an image from a monitoring camera that monitors the sky of a predetermined area as a processing target image and detecting an illegal flying object such as a drone. Further, the use of the image processing apparatus 1 is not limited to the use of detecting flying objects such as birds and drones. The image processing apparatus 1 can be used for the purpose of detecting an object other than a flying object such as an object moving on the ground and an object moving underwater.

  Further, examples of the pixel value acquired by the pixel value acquisition unit 13 include illuminance and hue, but which value is used may be changeable by the user of the image processing apparatus 1. For example, when the difference in brightness (difference in illuminance) between the background and the detection target object is not significant, the hue may be set to be used as the pixel value. On the other hand, when the color difference (hue difference) is not significant between the background and the detection target object, the illuminance may be set as the pixel value. Further, the pixel value acquisition unit 13 may acquire both illuminance and hue as pixel values, and the detection unit 14 may perform the above-described determination (variation determination, difference determination) for both illuminance and hue. In this case, for example, the detection unit 14 can detect the presence of the detection target object O when it can be determined that the detection target object O exists in the above-described determination for either illuminance or hue.

  Further, when comparing the magnitude relationship between two numerical values in the image processing apparatus 1, any of the two criteria “greater than” and “greater than” may be used, and “less than” and “less than” 2 Any of the two criteria may be used. The selection of such a standard does not change the technical significance of the process of comparing the magnitude relationship between two numerical values.

  DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus, 11 ... Center pixel determination part, 12 ... Peripheral pixel determination part, 13 ... Pixel value acquisition part, 14 ... Detection part, 15 ... Binarization part, C ... Center pixel, O ... Object to be detected, P: peripheral pixels, T: processing target image.

Claims (3)

A first pixel determining unit that determines a first pixel from a plurality of pixels in the processing target image;
A second pixel determining unit that determines a plurality of second pixels arranged around the first pixel apart from the first pixel in the processing target image;
A pixel value acquisition unit for acquiring the pixel value of the first pixel determined by the first pixel determination unit and the pixel value of each of the plurality of second pixels determined by the second pixel determination unit; ,
Based on the pixel value of the first pixel and the pixel value of each of the plurality of second pixels acquired by the pixel value acquisition unit, the object that does not overlap the plurality of second pixels and overlaps the first pixel and a detection unit that detects,
The detection unit performs a variation determination as to whether or not variation in pixel values of each of the plurality of second pixels is equal to or less than a predetermined reference, and in the variation determination, the variation is equal to or less than the predetermined reference. If determined, a difference determination is made as to whether or not a difference that is an absolute value of a difference between a representative value of each pixel value of the plurality of second pixels and a pixel value of the first pixel is equal to or greater than a predetermined threshold value. Run,
If it is determined in the difference determination that the difference is equal to or greater than the predetermined threshold, it is determined that the object exists,
If the variation is not determined to be less than or equal to the predetermined reference in the variation determination, or the difference is not determined to be greater than or equal to the predetermined threshold in the difference determination, it is determined that the object does not exist ,
Image processing device.   The image processing apparatus according to claim 1, wherein the second pixel determining unit determines the plurality of second pixels based on a size of an object that is predetermined as a detection target in the processing target image. A binarization unit that binarizes a pixel value of each pixel in the processing target image;
The first pixel determining unit sequentially determines each pixel in the processing target image as the first pixel,
Each process by the second pixel determination unit, the pixel value acquisition unit, and the detection unit is performed on each pixel determined as the first pixel by the first pixel determination unit,
In the binarization unit, the pixel value of the first pixel in which the object is detected by the detection unit and the pixel value of the first pixel in which the object is not detected by the detection unit are different from each other. binarizing the pixel values of each pixel in the process target image, the image processing apparatus according to claim 1 or 2.

Images