JP2019128797A - Attached matter detector and attached matter detection method - Google Patents

Abstract

To suppress omissions of attached matter detection.SOLUTION: An attached matter detector pertaining to an embodiment comprises a first extraction unit, a second extraction unit, and a detection unit. The first extraction unit extracts a first pixel group in which the luminance gradient of each pixel included in a captured image captured by an imaging device faces outward from a prescribed center region. The second extraction unit extracts a second pixel group in which the luminance gradient of each pixel faces inward toward the center region. The detection unit combines the first pixel group extracted by the first extraction unit and the second pixel group extracted by the second extraction unit and detects attached matter that adheres to the imaging device.SELECTED DRAWING: Figure 1B

Description

  The present invention relates to an attached matter detection device and an attached matter detection method.

  2. Description of the Related Art Conventionally, for example, there is an adhering matter detection device that detects an adhering matter adhering to a lens of a camera attached to a vehicle from a captured image taken by the camera (see, for example, Patent Document 1).

JP, 2010-14494, A

  However, in the prior art, when a water droplet is detected as a deposit, the water droplet can not be detected depending on the background image of the water droplet and the like, and there is room for improvement in the point of suppressing the detection leak of the deposit.

  The present invention has been made in view of the above, and it is an object of the present invention to provide a deposit detection apparatus and a deposit detection method capable of suppressing detection leak of the deposit.

  In order to solve the problems described above and achieve the object, the attached matter detection device according to the embodiment includes a first extraction unit, a second extraction unit, and a detection unit. The first extraction unit extracts a first pixel group in which the luminance gradient of each pixel included in the captured image captured by the imaging device is outward from a predetermined central region. The second extraction unit extracts a second pixel group in which the luminance gradient of each pixel is inward toward the central region. The detection unit combines the first pixel group extracted by the first extraction unit and the second pixel group extracted by the second extraction unit to detect an attached substance attached to the imaging device.

  ADVANTAGE OF THE INVENTION According to this invention, the detection leak of a deposit | attachment can be suppressed.

FIG. 1A is a diagram illustrating a mounting example of an attached matter detection apparatus. FIG. 1B is a diagram showing an outline of the deposit detection method. FIG. 2 is a block diagram of the attached matter detection device. FIG. 3A is a diagram (part 1) illustrating a specific example of processing by the conversion unit. FIG. 3B is a second diagram illustrating a specific example of processing performed by the conversion unit. FIG. 3C is a third diagram illustrating a specific example of the process performed by the conversion unit. FIG. 4A is a diagram showing a specific example of the first pixel group. FIG. 4B is a diagram showing a specific example of the second pixel group. FIG. 4C is a diagram illustrating a specific example of the arrangement order of the first pixel group. FIG. 5A is a diagram (part 1) illustrating a specific example of processing by the generation unit. FIG. 5B is a second diagram illustrating a specific example of processing performed by the generation unit. FIG. 6A is a diagram (part 1) illustrating a specific example of a process performed by the determination unit. FIG. 6B is a diagram (part 2) illustrating a specific example of processing by the determination unit. FIG. 7 is a diagram showing a specific example of processing by the determination unit. FIG. 8 is a flowchart showing a processing procedure performed by the attached matter detection device.

  Hereinafter, the attached matter detection device and the attached matter detection method according to the embodiment will be described in detail with reference to the attached drawings. Note that the present invention is not limited by the embodiments described below.

  First, the outline | summary of the deposit | attachment detection apparatus and deposit | attachment detection method which concern on embodiment is demonstrated using FIG. 1A and FIG. 1B. FIG. 1A is a diagram illustrating a mounting example of an attached matter detection apparatus. FIG. 1B is a diagram showing an outline of the deposit detection method.

  As shown in FIG. 1A, the deposit detection device 1 is mounted on a vehicle C. Further, the attached matter detection device 1 detects the attached matter F attached to the lens (not shown) of the camera 10 from the captured image L captured by the camera 10 which is an imaging device mounted on the vehicle C.

  The captured image L captured by the camera 10 is used for, for example, white line detection of the vehicle C and various types of sensing for automatic driving. In the example shown in the figure, a case where the vehicle C includes four cameras 10 that capture images in different directions is illustrated.

  Here, since the camera 10 is disposed outside the vehicle C, the lens F of the camera 10 may have the attached matter F such as water droplets or dust attached thereto.

  By the way, in the prior art, when detecting a water droplet as a deposit, the water droplet is detected from the captured image on the basis of one of a region brightened from the center of the droplet outward and a region darkened in the captured image.

  However, in the prior art, for example, such a water droplet may not be detected depending on a background image of the water droplet.

  Specifically, as shown in the captured image L of FIG. 1B, for example, when a stop line or the like is included in the background of the water droplet R, the water droplet R is darkened with a region where the brightness is brighter from the center to the outside And mixed regions. In the prior art, such water droplets cannot be detected, and there is a possibility that detection of water droplets may be missed.

  Therefore, in the attached matter detection method according to the embodiment, in the captured image L, a region in which the brightness is bright toward the outside and a region in which the brightness is dark toward the outside are respectively extracted, and the water droplets R are combined To detect.

  Specifically, as shown in FIG. 1B, in the attached matter detection method according to the embodiment, first, the first pixel group P1 and the second pixel group P2 are extracted from each pixel included in the captured image L (step S1) ). In the figure, the center point of each pixel is indicated by a black circle, and the direction of a line extending from the center point indicates the luminance gradient.

  For example, the first pixel group P1 is a group of pixels whose luminance gradient is directed outward from a predetermined central region, and the second pixel group P2 is a group of pixels whose luminance gradient is directed toward the central region. is there.

  That is, in the attached matter detection method according to the embodiment, the feature of the water droplet R brightened outward from the center is extracted as the first pixel group P1, and the feature of the water droplet R brightened inward toward the center is the second pixel Extract as group P2.

  Subsequently, in the attached matter detection method according to the embodiment, the attached matter F is detected based on the combination of the first pixel group P1 and the second pixel group P2 (step S2). For example, in the attached matter detection method, a region surrounded by the first pixel group P1 and the second pixel group P2 in the captured image L is detected as the attached matter F.

  That is, in the attached matter detection method, one water droplet R is detected by combining the feature of the water droplet R brightening outward from the center and the feature of the water droplet R brightening inward toward the center.

  This makes it possible to detect such water droplets R even in the case where a region where the brightness is bright from the center to the outside and a region where the brightness is dark are mixed in one water droplet R.

  Further, in the attached matter detection method according to the embodiment, a region surrounded by one of the first pixel group P1 and the second pixel group P2 is detected as a water droplet R whose brightness is brightened or darkened from the center to the outside. Is also possible.

  Therefore, according to the adhering matter detection method according to the embodiment, it is possible to suppress the detection omission of the adhering matter F.

  Next, the structure of the deposit | attachment detection apparatus 1 which concerns on embodiment is demonstrated using FIG. FIG. 2 is a block diagram of the attached matter detection device 1. In FIG. 2, the camera 10 and the vehicle control device 15 are shown together.

  The camera 10 includes an imaging element such as, for example, a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS), and images, for example, the surroundings of the vehicle C. The captured image L captured by the camera 10 is output to the attached matter detection apparatus 1.

  The vehicle control device 15 performs vehicle control of automatic driving of the vehicle C, automatic parking control, and driving assistance (for example, PCS (Pre-crash Safety System), AEB (Advanced Emergency Braking System), etc.). The vehicle control device 15 may be separate from an automatic parking control unit that performs automatic parking control.

  For example, the vehicle control device 15 can detect an obstacle or a white line from the captured image L input via the attached matter detection device 1, and can perform the vehicle control based on the detection result.

  The attached matter detection apparatus 1 includes a control unit 2 and a storage unit 3. The control unit 2 includes a conversion unit 21, a first extraction unit 22, a second extraction unit 23, a detection unit 24, and a determination unit 25. The control unit 2 includes, for example, a computer having a central processing unit (CPU), a read only memory (ROM), a random access memory (RAM), a hard disk drive (HDD), an input / output port, and various circuits.

  The CPU of the computer functions as, for example, the converter 21, the first extractor 22, the second extractor 23, the detector 24, and the determiner 25 of the control unit 2 by reading and executing the program stored in the ROM. To do.

  In addition, at least one or all or all of the conversion unit 21, the first extraction unit 22, the second extraction unit 23, the detection unit 24, and the determination unit 25 of the control unit 2 may be implemented as an application specific integrated circuit (ASIC) or an FPGA (field programmable) It can also be configured by hardware such as Gate Array).

  The storage unit 3 corresponds to, for example, a RAM or an HDD. The RAM and HDD can store generation condition information 31, determination condition information 32, score information 33, and information on various programs. In addition, the deposit | attachment detection apparatus 1 is good also as acquiring the above-mentioned program and various information via the other computer and portable recording medium which were connected by the wired or wireless network.

  The conversion unit 21 of the control unit 2 converts each pixel included in the captured image L into a code corresponding to the luminance gradient of each pixel. First, the conversion unit 21 converts the captured image L input from the camera 10 into a grayscale image by converting it to a grayscale. Gray scale conversion is a process of expressing each gradation from white to black according to the luminance of each pixel in the captured image L.

  Subsequently, the conversion unit 21 extracts edge intensities in the X-axis direction and the Y-axis direction of each pixel by using a Sobel filter on the gray scale image. Next, the conversion unit 21 calculates a luminance gradient from each edge strength, and converts each pixel into a corresponding code.

  3A to 3C are diagrams illustrating specific examples of processing performed by the conversion unit 21. FIG. 3A shows edge strengths in the X-axis direction and the Y-axis direction. The conversion unit 21 calculates the actual edge strength S and the edge direction θ from the edge strength in the X-axis direction and the edge strength in the Y-axis direction for each pixel.

  As shown in FIG. 3A, the diagonal of the rectangle having edge strength in the X-axis direction and edge strength in the Y-axis direction as one side is the edge strength S, and the angle between the diagonal and the X-axis is the edge direction θ It becomes. The direction θ of the edge is a brightness gradient.

  In the present embodiment, the conversion unit 21 calculates a luminance gradient by integrating a plurality of pixels. In other words, the conversion unit 21 divides the captured image L into a plurality of regions and converts the images into codes corresponding to the luminance gradient for each region.

  FIG. 3B is a diagram illustrating a specific example of pixels to be integrated. The conversion unit 21 calculates the luminance gradient of the cell of interest located at the center of each block shown in FIG. 3B. Here, the block is, for example, a set of pixels of 3 × 3 cells, and a cell is a set of 4 × 4 pixels (pixels).

  That is, in the example shown in the figure, one block is a set of 12 × 12 pixels. Here, the conversion unit 21 calculates a representative value of the luminance gradient of the target cell for each block. Specifically, as illustrated in FIG. 3C, the conversion unit 21 creates a histogram for each block, and calculates a representative value of the cell of interest based on the histogram.

  Specifically, as shown in FIG. 3C, the conversion unit 21 classifies the luminance gradient into a plurality of classes at predetermined intervals, and sets the edge intensity of each pixel to a class that matches the luminance gradient of each pixel included in the block. Add. In the example shown in the figure, the case is shown where the class is classified every 20 °.

  Then, the conversion unit 21 calculates the class having the highest sum of the edge strengths as the luminance gradient of the target cell. At this time, when the sum is equal to or greater than a threshold, the conversion unit 21 sets the class as a representative value. In other words, a representative value is not assigned to a cell of interest for which the sum is equal to or less than the threshold.

  That is, when the edge strength of each pixel included in the block is low, or when there is a variation in the luminance gradient of each pixel, a representative value is not allocated to the target cell.

  When the conversion unit 21 calculates a representative value for one target cell, the conversion unit 21 continues to shift the block by one cell to set the target cell, and calculates a representative value for the target cell. Thereby, the conversion unit 21 calculates a representative value for each cell.

  Then, the conversion unit 21 converts each cell into a code of a class corresponding to the representative value. As a result, in the encoded captured image L, the codes are arranged in a lattice. The captured image L is output to the first extraction unit 22.

  Returning to the description of FIG. 2, the first extraction unit 22 will be described. The first extraction unit 22 extracts a first pixel group P1 in which the luminance gradient of each pixel included in the captured image L is outward from a predetermined center region. The second extraction unit 23 extracts a second pixel group P2 in which the luminance gradient of each pixel is inward toward the central region.

  The first extraction unit 22 and the second extraction unit 23 use the first pixel group P1 and the second pixel group P2 to generate code strings that satisfy a predetermined arrangement order from the captured image L encoded by the conversion unit 21 according to the luminance gradient. Respectively.

  Here, the specific example of the process by the 1st extraction part 22 and the 2nd extraction part 23 is demonstrated using FIG. 4A-FIG. 4C. FIG. 4A is a diagram illustrating a specific example of the first pixel group P1. FIG. 4B is a diagram illustrating a specific example of the second pixel group P2. FIG. 4C is a diagram illustrating a specific example of the arrangement order of the first pixel group P1.

  First, the first pixel group P1 and the second pixel group P2 will be described using FIGS. 4A and 4B. In FIG. 4A and FIG. 4B, actual edge orientations (brightness gradients) are schematically shown in place of reference numerals for the purpose of visual intelligibility.

  The first extraction unit 22 extracts, from the captured image L, code strings that satisfy the arrangement order of the upper side pattern Pu1, the lower side pattern Pd1, the left side pattern Pl1, and the right side pattern Pr1 as a first pixel group P1.

  The upper side pattern Pu1 indicates the arrangement sequence of the code string corresponding to the upper part of the water droplet that becomes brighter from the center toward the end. The lower side pattern Pd1 indicates the arrangement order of the code string corresponding to the lower part of the water droplet. Further, the left side pattern Pl1 and the right side pattern Pr1 indicate the arrangement order of the code strings corresponding to the left side and the right side of the water droplet, respectively.

  On the other hand, as shown in FIG. 4B, the second extraction unit 23 sets a code string satisfying the arrangement order of the upper side pattern Pu2, the lower side pattern Pd2, the left side pattern Pl2 and the right side pattern Pr2 from the captured image L as the second pixel group P2. Extract.

  The upper side pattern Pu2 indicates the arrangement sequence of the code string corresponding to the upper part of the water droplet that becomes darker from the center toward the end. The lower side pattern Pd2 indicates the arrangement order of the code string corresponding to the lower part of the water droplet. Further, the left side pattern Pl2 and the right side pattern Pr2 indicate the arrangement order of the code strings corresponding to the left side and the right side of the water droplet, respectively.

  As described above, the first extraction unit 22 extracts the pixel group showing the feature of the water droplet which is brightened from the center to the end as the first pixel group P1, and the second extraction unit 23 proceeds from the center to the end A pixel group that becomes dark is extracted as a second pixel group P2.

  Incidentally, the first pixel group P1 and the second pixel group P2 have different code string lengths depending on the size of the water droplets. For example, with a small water droplet, the length of the code string on each side is shorter than with a large water droplet.

  Therefore, the first extraction unit 22 and the second extraction unit 23 allow repetition to extract the first pixel group P1 and the second pixel group P2, as long as the above arrangement order is satisfied.

  Specifically, as shown in FIG. 4C, when the upper side pattern Pu1 of the first pixel group P1 is represented by each code string (A to F), as shown in FIG. 4A, the patterns are arranged in the order of A to F. If so, repeat of each code is permitted.

  That is, the first extraction unit 22 can extract the upper side pattern Pu1 if it satisfies the arrangement order such that A or B is to the right of A and B or C is to the right of B.

  This makes it possible to extract water droplets of different sizes by one extraction process. Therefore, since the deposit | attachment detection apparatus 1 which concerns on embodiment can extract the several water droplet from which a magnitude | size differs by one process, it can suppress the detection leak of a water droplet, suppressing process load.

  In addition, since water drops are generally spherical, the number of repetitions of each symbol should be axisymmetric from the center. For this reason, the deposit | attachment detection apparatus 1 which concerns on embodiment can also exclude a code sequence with a bad balance in the code sequences which satisfy | fill arrangement | sequence order.

  As shown in FIG. 6B, for example, the balance between A and F located at both ends is scrutinized. Here, in the same figure, the case where A is repeated 3 times and F is repeated 10 times is shown.

  At this time, in the case where the numbers of A and F differ by 2 times or more, the attached matter detection device 1 excludes such a code string even if the arrangement order is satisfied. Thereby, the detection accuracy of a water droplet can be improved. In other words, detection of a subject other than the water droplet as the water droplet can be suppressed.

  At this time, as shown in FIG. 3C, the adhering matter detection apparatus 1 may extract the code string so that the balance is uniform. Specifically, only three Fs from the center may be detected as the characteristics of the water droplets, and the fourth and later may be excluded from the center and extracted. Although the upper side pattern Pu1 has been described as an example here, the same applies to other patterns.

  Returning to the description of FIG. 2, the detection unit 24 will be described. The detection unit 24 detects one attached substance based on a combination of the first pixel group P1 extracted by the first extraction unit 22 and the second pixel group P2 extracted by the second extraction unit 23.

  The detection unit 24 includes a generation unit 24a and a determination unit 24b. The generation unit 24a generates an integrated area in which a first area extending from the first pixel group P1 to the central area and a second area extending from the second pixel group P2 to the central area are integrated.

  The determination unit 24b determines the integrated region as the adhering matter based on the combination of the first pixel group P1 and the second pixel group P2 that configure the integrated region generated by the generation unit 24a.

  Here, as described above, the detection unit 24 can detect a water droplet that becomes bright from the center to the end, a water droplet that becomes dark from the center to the end, and a water droplet having both features.

  In the present embodiment, a water droplet constituted only by the first pixel group P1 is a water droplet which becomes bright from the center to the outside, and a water droplet constituted only by the second pixel group P2 is dark from the center to the outside It is a water drop.

  Further, the water droplet constituted by the first pixel group P1 and the second pixel group P2 is a water droplet having characteristics of a water droplet brightening from the center to the outside and a water droplet darkening from the center to the outside.

  The detection unit 24 detects a water droplet including the first pixel group P1 and the second pixel group P2 with detection conditions strengthened more than a water droplet formed of one of the first pixel group P1 or the second pixel group P2. . That is, for the water droplet including the first pixel group P1 and the second pixel group P2, the detection unit 24 includes the water droplet including the first pixel group P1 and the first pixel group P1, and the second pixel group P2. Detection is performed with a stronger detection condition than a water droplet composed of the second pixel group P2.

  This is because a water droplet including the first pixel group P1 and the second pixel group P2 is more erroneously detected than a water droplet formed of one of the first pixel group P1 or the second pixel group P2, that is, an object other than the water droplet Is likely to be detected incorrectly as a water drop.

  Specifically, the false detection is suppressed by strengthening the generation condition of the integrated region by the generation unit 24a and the determination condition of the water droplet by the determination unit 24b. Such generation conditions are stored as generation condition information 31 in the storage unit 3, and such determination conditions are stored as determination condition information 32 in the storage unit 3.

  FIG. 5A and FIG. 5B are diagrams showing specific examples of processing by the generation unit 24a. First, an integrated region Ri based on the first pixel group P1 will be described using FIG. 5A.

  As illustrated in FIG. 5A, the generation unit 24a integrates the first area Ru1 corresponding to the upper side pattern Pu1 of the first pixel group P1 and the first area Rd1 corresponding to the lower side pattern Pd1 of the first pixel group P1. Thus, a single area Rio which is an integrated area Ri is generated.

  The first region Ru1 is a square region extending from the upper side pattern Pu1 to the center (downward in the figure), with the length of the upper side pattern Pu1 as one side, and the first region Rd1 is a long side of the lower side pattern Pd1. This is a square region having one side and extending from the lower side pattern Pd1 to the center (upward in the figure).

  At this time, the generation unit 24a sets the overlapping condition of the first region Ru1 and the first region Rd1 to be 50% or more as an integration condition of the single region Rio. Further, as shown in the figure, the single region Rio is all the regions including the first region Ru1 and the first region Ru2. That is, the single region Rio is a logical sum of the first region Ru1 and the first region Ru2.

  On the other hand, as shown in FIG. 5B, with regard to the mixed area Rim which is the integrated area Ri including the first pixel group P1 and the second pixel group P2, the integration condition is strengthened more than the single area Rio.

  Specifically, for the mixed region Rim, the overlapping ratio between the first region Ru1 corresponding to the upper side pattern Pu1 of the first pixel group P1 and the second region Rd2 corresponding to the lower side pattern Pd2 of the second pixel group P2 is The integration condition is 90% or more.

  Further, as shown in the figure, the generation unit 24a generates, as a mixed region Rim, a region where the first region Ru1 and the second region Rd2 overlap. That is, the mixed region Rim is a logical product of the first region Ru1 and the second region Rd2.

  Subsequently, a specific example of the process by the determination unit 24b will be described with reference to FIGS. 6A and 6B. FIG. 6A and FIG. 6B are diagrams showing a specific example of processing by the determination unit 24b.

  First, the determination condition in the case of determining the deposit as the single region Rio will be described with reference to FIG. 6A. As shown in the figure, for example, when a pattern having three or more sides is included in a single area Rio, the determination unit 24b determines such a single area Rio as a deposit area Fo to which a deposit is attached. To do.

  Specifically, the determination conditions in the single region Rio are the first condition in which the upper side pattern Pu1 exists in the upper side, the second condition in which the lower side pattern Pd1 exists in the lower side, and the first condition in which the left side pattern Pl1 exists in the left side. In the case where the third condition is the fourth condition in which the right side pattern Pr1 exists in the right side portion, at least three conditions are satisfied.

  On the other hand, as illustrated in FIG. 6B, the determination unit 24b determines that the pattern of each side including the mixed area Rim is four sides in the mixed area Rim as the determination condition of the attached substance area Fm.

  That is, the first condition that one of the upper side patterns Pu1 and Pu2 exists in the upper side, the second condition that one of the lower side patterns Pd1 and Pd2 exists in the lower side, and the one that one of the left side patterns Pl1 and Pl2 exists in the left side A mixed region Rim that satisfies all the fourth conditions in which one of the three conditions and one of the right side patterns Pr1 and Pr2 is present in the right side portion is determined as the attached matter region Fm.

  The attached matter area Fo is an area formed by one of the first pixel group P1 and the second pixel group P2, and the adhered matter area Fm is formed by a combination of the first pixel group P1 and the second pixel group P2. This is the area that has been

  As described above, the generation unit 24a can suppress erroneous detection of water droplets in the mixed area Rim by the determination unit 24b by strengthening the integration condition of the mixed area Rim more than the mixing condition of the single area Rio.

  Moreover, the determination part 24b can suppress the misdetection of a water droplet by strengthening the determination conditions of the deposit | attachment area | region Fm with respect to the mixing area | region Rim rather than the deposit | attachment area | region Fo with respect to the single area | region Rio.

  As described above, the detection unit 24 reinforces the conditions for the water droplets including the first pixel group P1 and the second pixel group P2 as compared to the water droplets configured from one of the first pixel group P1 or the second pixel group P2. To detect.

  Thereby, the deposit | attachment detection apparatus 1 which concerns on embodiment can suppress the detection leak of a water droplet, suppressing the misdetection of a water droplet.

  Moreover, the deposit | attachment detection apparatus 1 which concerns on embodiment detects a water drop not by catching the outline part of a circular water droplet but by catching the characteristic of each side of the rectangle included in a water drop. This makes it possible to detect water droplets of various shapes in a single process, regardless of the perfect circular shape or the elliptical shape.

  Returning to the description of FIG. 2, the determination unit 25 will be described. The determination unit 25 determines the attachment regions Fo and Fm based on the time-series overlap of the integrated regions Ri determined as the attachment regions Fo and Fm by the determination unit 24b.

  FIG. 7 is a diagram illustrating a specific example of the process performed by the determination unit 25. As illustrated in FIG. 7, the determination unit 25 adds points to the attached matter area Fo or the area detected as the attached matter area Fm, and deducts points from the area not detected as the attached matter area Fo or the attached matter area Fm.

  Then, the determination unit 25 determines the region where the accumulated value is equal to or more than the threshold value as the attached matter region. That is, the area continuously detected as the deposit area Fm or the deposit area Fo is determined as the deposit area.

  As described above, the determination unit 25 can improve the detection accuracy of the attached region by determining the attached region based on the plurality of captured images L that are continuous in time series. That is, since the determination unit 25 does not determine only one captured image L as an attached matter region, it is possible to suppress erroneous detection of the attached matter.

  Here, the determination unit 25 is an appendage configured by a combination of the first pixel group P1 and the second pixel group P2 with respect to the deposit region Fo configured by one of the first pixel group P1 or the second pixel group P2. The weight of the kimono area Fm is reduced to determine the deposit area.

  Specifically, the determination unit 25 sets the score for the attached matter area Fm lower than that of the attached matter area Fo to determine the attached matter area. For example, when the deposit region Fo is detected five times, the deposit region is determined as the deposit region when the deposit region Fm is detected ten times.

  Thus, the determination unit 25 can suppress erroneous detection based on the attached matter region Fm by changing the weighting between the attached matter region Fo and the attached matter region Fm to determine the attached matter region. The weighting of the attached matter area Fo and the attached matter area Fm can be arbitrarily changed.

  Further, when the determination unit 25 determines the attachment region, the determination unit 25 generates a mask image in which the attachment region is masked. And the determination part 25 superimposes a mask image on the captured image L, and outputs it to the vehicle control apparatus 15 (refer FIG. 2).

  Thereby, the vehicle control apparatus 15 can suppress various erroneous sensings due to the deposits. That is, it is possible to suppress erroneous vehicle control due to the attached matter. For example, the vehicle control device 15 stops the automatic parking control when it is determined that it is difficult to continue the automatic parking control safely because the area of the deposit in the captured image L is large. Thus, the vehicle control device 15 can improve the safety of the vehicle by performing the automatic parking control only when the safety of the vehicle is ensured.

  Next, a processing procedure performed by the attached matter detection device 1 according to the embodiment will be described with reference to FIG. FIG. 8 is a flowchart showing a processing procedure performed by the attached matter detection device 1.

  As shown in FIG. 8, first, the conversion unit 21 converts the captured image L (step S101), and the first extraction unit 22 and the second extraction unit 23 convert the first pixel group P1 and the second pixel group P2. Extract (step S102).

  Subsequently, the generation unit 24a generates an integrated region Ri (step S103). Subsequently, the determination unit 24b determines whether or not the integrated region Ri is the mixed region Rim (step S104).

  Here, when the integrated region Ri is the mixed region Rim (Yes at step S104), the determination unit 24b determines whether the mixed region Rim is configured by four sides (step S105).

  Then, when the mixed region Rim includes four sides (Yes at Step S105), the determination unit 24b determines the mixed region Rim as an attachment region (Step S106).

  On the other hand, when the mixed area Rim has three or less sides (No at step S105), the determining unit 24b omits the process at step S106. In addition, when the integrated region Ri is not the mixed region Rim (No in step S104), the determining unit 24b determines whether the single region Rio is configured with three or more sides because the integrated region Ri is a single region Rio. Is determined (step S110).

  Here, when the single area Rio is configured with three or more sides (Yes in step S110), the determination unit 24b proceeds to the process of step S106, and the single area Rio is configured with two or less sides. If it is (Step S110, No), the process proceeds to Step S107.

  Subsequently, the determination unit 25 adds or deducts a score for each area based on the attached matter area determined by the determination unit 24b (step S107). Then, the determination unit 25 determines whether the total value of each area is larger than a threshold (step S108).

  Here, when the cumulative value is larger than the threshold value (step S108, Yes), the determination unit 25 determines the attached region (step S109) and ends the process. On the other hand, when the total value is equal to or less than the threshold (No in step S108), the determination unit 25 omits the process of step S109 and ends the process.

  As described above, the attached matter detection device 1 according to the embodiment includes the first extraction unit 22, the second extraction unit 23, and the detection unit 24. The first extraction unit 22 extracts a first pixel group P <b> 1 in which the luminance gradient of each pixel included in the captured image L captured by the camera 10 (an example of the imaging device) is outward from a predetermined central region. The second extraction unit 23 extracts a second pixel group P2 in which the luminance gradient of each pixel is inward toward the central region. The detection unit 24 detects one attached matter attached to the camera 10 by combining the first pixel group P1 extracted by the first extraction unit 22 and the second pixel group P2 extracted by the second extraction unit 23. . Therefore, according to the deposit | attachment detection apparatus 1 which concerns on embodiment, the detection leak of a deposit | attachment can be suppressed.

  By the way, although the above-mentioned embodiment showed the case where all the adhesion detection devices 1 were applied to camera 10 for vehicles, for example, other things, such as surveillance / crime prevention camera etc. which are set up in the outside of a building, alley It may be applied to different types of cameras.

  Moreover, although embodiment mentioned above demonstrated the case where the adhering matter detection apparatus 1 detected a water droplet as an adhering matter, it is not limited to this. That is, it is also possible to apply to other deposits instead of the water droplets.

  Further effects and modifications can be easily derived by those skilled in the art. Thus, the broader aspects of the present invention are not limited to the specific details and representative embodiments shown and described above. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents.

DESCRIPTION OF SYMBOLS 1 Adherence detection apparatus 21 Conversion part 22 1st extraction part 23 2nd extraction part 24 Detection part 24a Generation part 24b Determination part 25 Determination part P1 1st pixel group P2 2nd pixel group Ri Integrated area Rio Single area | region Rim Mixed area | region

Claims (7)

A first extraction unit that extracts a first pixel group in which a luminance gradient of each pixel included in a captured image captured by the imaging device is outward from a predetermined center region;
A second extraction unit that extracts a second pixel group in which the luminance gradient of each pixel is directed toward the central region;
A detection unit configured to detect the adhering matter attached to the imaging device by combining the first pixel group extracted by the first extraction unit and the second pixel group extracted by the second extraction unit. An attached matter detection device characterized by The detection unit is
A generating unit that generates an integrated region in which a first region extending from the first pixel group to the central region and a second region extending from the second pixel group to the central region are integrated;
And a determining unit that determines the integrated region as the deposit based on a combination of the first pixel group that forms the integrated region generated by the generating unit and the second pixel group. The attached matter detection device according to claim 1. The generator is
When the integrated region is generated by integrating the first region and the second region, the integrated region is formed by integrating the first region and the first region or the second region and the second region. The adhesion detection device according to claim 2, wherein integration conditions are strengthened more than generation. The determination unit
For the integrated region generated by integrating the first region and the second region, the integrated region or the second region generated by integrating the first region and the first region and the first region The deposit detection apparatus according to claim 2 or 3, wherein the deposit determination condition is reinforced more than the integrated region generated by integrating two regions. A determination unit for determining the deposit based on a time-series overlap of the integrated region determined as the deposit by the determination unit;
The determining unit is
For the integrated region generated by integrating the first region and the second region, the integrated region or the second region generated by integrating the first region and the first region and the first region The adhering matter detection device according to claim 2, 3 or 4, wherein the weight is lighter than that of the integrated region generated by integrating the two regions. The imaging device is mounted on a vehicle,
The vehicle further includes a parking control unit that performs parking control of the vehicle based on the captured image.
The parking control unit
The adhering matter detection apparatus according to claim 5, wherein when the adhering matter is confirmed by the determining unit, parking control of the vehicle is stopped. A first extraction step of extracting a first pixel group in which a luminance gradient of each pixel included in a captured image captured by the imaging device is outward from a predetermined center region;
A second extraction step of extracting a second pixel group in which the luminance gradient of each pixel is directed toward the central region;
And a detection step of detecting an adhering matter attached to the imaging device by combining the first pixel group extracted by the first extraction step and the second pixel group extracted by the second extraction step. An attached matter detection method characterized by

Images