JP4699773B2 - Speed sensor, speed sensor program, and mobile body equipped with these - Google Patents

Description

  The present invention relates to a speed sensor and a speed sensor program, and more particularly to a speed sensor and a speed sensor program suitable for spraying agricultural chemicals and fertilizer in conjunction with the speed of a tractor in a field.

  Conventionally, a method for detecting the speed of a tractor using a proximity sensor is known for a fertilizer spreader or the like that performs work in conjunction with the speed of a tractor (Patent Document 1). In this method, a detected object such as a magnet is provided on a tire wheel of a grounded tire, and a proximity sensor such as a magnetic sensor is attached to the tractor body side. And the rotation amount of a tire wheel is acquired by detecting a to-be-detected body with this proximity sensor, and the speed of a tractor is derived | led-out based on the moving distance calculated from this rotation amount.

  In addition to the proximity sensor described above, as a method using a non-contact type sensor, there is known a method for detecting a speed using a Doppler effect caused by reflection of radar, ultrasonic waves, or the like (Patent Document 2). In this method, transmitting means for transmitting laser light, microwaves, ultrasonic waves, and the like, and receiving means for receiving reflected waves transmitted from the transmitting means and reflected by the ground are attached to the tractor body. Then, the speed is measured based on the change in the frequency of the received wave with respect to the transmitted wave. Furthermore, in recent years, a method for detecting speed using a GPS (Global Positioning System) signal is known.

Japanese Utility Model Publication No. 2-120121 JP 2001-306107 A

  However, in the method using the proximity sensor including the invention described in Patent Document 1, the change in the effective diameter due to the slip of the tire with respect to the contact surface and the load condition on the tire greatly affects the calculated travel distance. Therefore, there is a problem that the actual moving speed cannot be derived with high accuracy. Especially in agricultural tractors, the impact is great because it is not paved ground. In addition, since the size of tires and usage conditions differ for each work machine, there is also the inconvenience of having to install a proximity sensor according to various conditions.

  In order to confirm how much slip actually occurs in the field, the present applicant measured the slip ratio with respect to the ground of the tire when the traveling speed and the weight of the aircraft were changed using a beat harvester. There is a fact that tires slip at an average rate of about 10%. Therefore, when the working machine is a pesticide spraying machine (sprayer), the pesticide is sprayed in conjunction with the speed, and therefore the spraying amount of the pesticide is not uniform when the speed deviates by 10%. When the speed sensor number is calculated 10% more than the actual speed of the agricultural chemical spraying machine as in the above-described experiment, agricultural chemicals are excessively sprayed, which may cause damage to the crops. Further, in the fertilizer application work machine for applying fertilizer, the same problem occurs when the application amount is controlled in conjunction with the speed data. Or, when using the speed data with a self-propelled harvester and trying to control the rotation speed of the harvester's digging conveyor, there is also a problem that the harvest will be damaged if the expected speed data is not obtained .

The slip ratio (%) was determined according to the following definition.
Slip rate = 100 * (distance to be originally traveled−actual distance traveled) / (distance to be originally traveled)
Original distance to travel: Value obtained by multiplying the circumference of the tire by the number of revolutions Actual distance traveled: Distance the work implement actually traveled on the field

  In addition, the method using radar, ultrasonic waves, etc., including the invention described in Patent Document 2, is extremely expensive and difficult to put into practical use. Since permission is required, the procedure is complicated. In particular, ultrasonic waves are easily affected by disturbances such as miscellaneous signals, and there is a problem that sufficient detection accuracy cannot be obtained in an environment such as a field. Also, in the method using the GPS signal, if the GPS signal is interrupted by an obstacle such as a windbreak forest around the field, the speed may not be detected, and this method also has a high cost for the apparatus. There's a problem.

  The present invention has been made to solve such problems, and is a speed at which the actual speed with respect to the ground of an agricultural tractor or the like in a field can be measured inexpensively and accurately without being affected by tire slip or the like. It aims to provide a sensor and speed sensor program.

The speed sensor according to the present invention is characterized by an imaging unit that images the ground from a moving body, a storage unit that stores a captured image captured by the imaging unit, and a single still image obtained from the storage unit. A speed sensor comprising: an arithmetic processing unit that cuts out a predetermined template image and obtains a still image after a predetermined time has elapsed and collates with the template image to obtain the number of moving pixels, thereby calculating the speed of the moving body. The arithmetic processing means acquires a reference image serving as a reference for speed calculation, a still image acquisition unit that acquires a collation target image to be collated after a predetermined time, and a template from within the reference image. The image is cut out, and the maximum brightness gradient direction of the pixels constituting the image is acquired as a code, and a search area within a predetermined range from the verification target image is obtained. The direction encoding processing unit that acquires the maximum brightness gradient direction of the pixels constituting the search area as a code by comparing the code of the maximum gradient direction of the template image with the code of the maximum gradient direction of the search area and obtains the sum of the residual absolute value and the direction code collating unit for collating with the average residual absolute value obtained by dividing the sum by the total number of pixels, the result of collation, the average residual absolute in the search area the same similar image specifying unit value to identify the smallest image as the same similar image of the template image, and the template image and the same similar image moving pixel number to calculate the number of pixels moved by comparing the calculation unit, was calculated And an actual speed calculation unit that calculates an actual speed of the moving body based on the number of moving pixels.

  In the present invention, the direction encoding processing unit or the moving pixel number calculating unit cuts out a plurality of template images from one reference image and executes direction encoding processing, and is the same as each template image or While calculating the number of moving pixels by obtaining similar identical images, the arithmetic processing means obtains a difference in the number of moving pixels in the plurality of template images, and whether or not the difference is equal to or less than a predetermined threshold value. Whether or not the similarity based on the direction code between the template image and the same similar image is less than a predetermined threshold when both the moving pixel number difference determination unit and the moving pixel number difference are not less than the threshold A determination unit for determining the difference between the moving pixel numbers based on the determination results of the determining units. If there is a template image that is less than or equal to a certain threshold value, the average number of moving pixels is calculated from the number of those moving pixels. On the other hand, the difference in the number of moving pixels is not less than or equal to a predetermined threshold value. As a result, when there is a template image whose similarity is equal to or less than a predetermined threshold and the same similar image, it is preferable to calculate the number of moving pixels. According to such a configuration, the actual speed can be calculated based on the number of moving pixels with a small error, and the accuracy can be improved.

  Furthermore, in the present invention, when the direction code collating unit compares the direction codes between the pixels to obtain the sum of the residual absolute values, the direction code collating unit performs a comparison with the sum of the previous residual absolute values, and more than the sum If it becomes larger, it is preferable to stop the calculation process at that time and shift to the collation process with the next image in the search area. According to such a configuration, the load of the collation process is reduced and the speed calculation process is accelerated.

In the present invention, it is preferable that the arithmetic processing unit includes a speed range setting unit that can arbitrarily set a time interval from the acquisition of the reference image to the acquisition of the verification target image according to the speed of the moving body. According to such a configuration, since the image acquisition time interval is determined according to the actual speed, it is possible to prevent a problem that the same similar image that approximates the template image is framed out from the collation target image. The actual speed can be calculated reliably. Further, in the present invention, when the direction encoding processing unit cuts out three template images from one reference image, and when the second template image is cut out from the center position of the reference image, the moving pixel The number difference determination unit calculates a difference in the number of moving pixels between the second template image and the first template image, and a difference in the number of moving pixels between the second template image and the third template image. It may be determined whether or not is below a predetermined threshold.

Further, the speed sensor program according to the present invention is characterized in that a storage unit that stores a photographed image of the ground photographed from a moving body by the photographing unit, and a predetermined template image obtained by acquiring one still image from the storage unit. And a processing unit for calculating the speed of the moving object by obtaining a still image after a predetermined time and collating with the template image to obtain the number of moving pixels, and a display for displaying the calculated speed A speed sensor program for causing a computer to execute as a means, wherein the arithmetic processing means obtains a reference image serving as a speed calculation reference from the storage means, and extracts a template image from the reference image. A template image direction encoding step for obtaining, as a code, the maximum brightness gradient direction of the pixels constituting A collation target image acquisition step of acquiring a collation target image at a predetermined time interval after acquiring the reference image, and a maximum brightness of pixels constituting the search region by selecting a search region within a predetermined range from the collation target image A search region direction encoding step for obtaining a gradient direction as a sign, and a sum of absolute values of residuals obtained by comparing a sign of the maximum gradient direction of the template image and a sign of the maximum gradient direction of the search area is obtained, and this sum is obtained. A direction code collation step for collation using an average residual absolute value divided by the total number of pixels, and an image having the smallest average residual absolute value in the search region as a result of the collation is the same similar image of the template image same similar images and specifying step, the template image and the number of pixels moved calculated stearyl for calculating the number of pixels moved by comparing the same similar image identifying as And flop is the real speed calculating step of calculating the actual speed of the moving object based on the moving number of pixels the calculated point to be executed by a computer.

  Further, in the present invention, in the template image direction encoding step or the moving pixel number calculating step, a plurality of template images are cut out from one reference image and a direction encoding process is executed, which is the same as each template image. Alternatively, the number of moving pixels is calculated by obtaining similar identical images, and further, a difference in the number of moving pixels in the plurality of template images is obtained, and it is determined whether or not the difference is equal to or less than a predetermined threshold value. If there is a template image in which the difference in the number of moving pixels is equal to or less than a predetermined threshold as a result of the difference in the number of moving pixels and the determination of the difference in the number of moving pixels, the average number of moving pixels is calculated from the number of moving pixels. As a result of the average moving pixel number calculating step and the determination of the moving pixel number difference, if the moving pixel number difference is not less than the threshold value, the test A similarity determination step for determining whether or not the similarity based on the direction code between the plate image and the same similar image is equal to or less than a predetermined threshold, and as a result of the similarity determination, the similarity is equal to or less than a predetermined threshold If the template image and the same similar image are present, it is preferable to cause the computer to execute a similarity-based moving pixel number calculating step for calculating the moving pixel number.

  Further, in the present invention, in the direction code collation step, when comparing the direction codes between the pixels and obtaining the sum of the residual absolute values, the sum of the residual absolute values is compared, If it becomes larger, it is preferable to stop the calculation process at that time and shift to the collation process with the next image in the search area.

Further, in the present invention, when the speed range of the moving object is changed by the speed range setting unit, a speed range setting step for setting a time interval from acquisition of the reference image to acquisition of the comparison target image according to the speed range is executed. It is preferable to make it. Further, in the present invention, in the template image direction encoding step, when three template images are cut out from one reference image, and the second template image is cut out from the center position of the reference image, the moving pixel In the number difference determination step, a difference in the number of moving pixels between the second template image and the first template image and a difference in the number of moving pixels between the second template image and the third template image are calculated, and each difference is calculated. It may be determined whether or not is below a predetermined threshold. Moreover, the mobile body which concerns on this invention is provided with the speed sensor in any one of Claims 1-4, or the speed sensor program in any one of Claims 5-8.

  According to the present invention, it is possible to measure the actual speed of an agricultural tractor or the like in the field without being affected by the slip of the tire, and in conjunction with the speed of the agricultural tractor, pesticide spraying, fertilizer spraying, crop harvesting, and the like are performed. The work can be performed with high accuracy.

  Hereinafter, embodiments of a speed sensor and a speed sensor program according to the present invention will be described with reference to the drawings.

  FIG. 1 is a configuration block diagram showing the configuration of the speed sensor 1 in the present embodiment, and FIG. 2 is a diagram showing the speed measurement principle of the present embodiment. The speed sensor 1 according to the present embodiment is mounted on a moving body such as a tractor, and mainly includes a photographing unit 2 configured by a CCD camera or the like that photographs the ground from the moving body, and light emission that illuminates a photographing region on the ground. The illumination means 3 is composed of a diode and the like, and is composed of a RAM (Random Access Memory), a ROM (Read Only Memory), and the like that store various information including a photographed image taken by the photographing means 2 and a speed sensor program. Storage means 4, calculation processing means 5 for calculating the speed of the moving body, speed display means 6 including a liquid crystal display for displaying the speed, and input means 7 for inputting various settings. is doing.

  As shown in FIG. 2, the speed measurement principle in this embodiment is based on detecting the actual moving distance in the field as the number of moving pixels in the image taken by the photographing means 2 such as a camera. The speed is derived from the relationship of travel time. Here, when the height H of the camera lens from the field and the focal length f of the camera are used, the actual speed V of the moving body is in a relationship of V = (H / f) V ′. Here, since the relative speed V ′ is a value obtained by the number of moving pixels and time, it can be seen that the actual speed V is determined by the ratio of H and f.

Next, each constituent unit constituting the present embodiment will be described.
The photographing means 2 is fixed to the moving body so that the ground can be photographed from as high a height as possible, and photographs and sequentially records moving images. The captured moving image may be transmitted to the storage unit 4 or stored in a storage unit (not shown) built in the CCD camera, and then only necessary still images are acquired and transmitted to the storage unit 4. It may be. An illumination unit 3 is disposed in the vicinity of the photographing unit 2 so that the photographing region can be illuminated accurately. By illuminating the shooting area with such illumination means 3, the change in brightness due to the weather and shade is suppressed, so that the brightness difference in the still image can be acquired more accurately.

  The arithmetic processing means 5 is composed of a CPU (Central Processing Unit) and the like, and controls each component based on a speed sensor program stored in the storage means 4 and acquires various data and setting information. The necessary processing is executed as needed. The main processing of the arithmetic processing unit 5 will be described. One sample reference image is acquired from the photographed image stored in the storage unit 4, a template image is cut out from a predetermined position, and a predetermined time has elapsed. A collation target image is acquired as another still image later, the template image and the collation target image are collated to obtain an image as similar as possible to the template image, and the number of moving pixels of the template image in a predetermined time And the actual speed of the moving object is calculated based on the number of moving pixels.

  Next, the configuration of the arithmetic processing means 5 that performs the processing as described above will be described. The arithmetic processing unit 5 mainly includes a speed range setting unit 51, a still image acquisition unit 52, a direction encoding processing unit 53, a direction code matching unit 54, an identical similar image specifying unit 55, and a moving pixel number calculating unit. 56, a moving pixel number difference determination unit 57, a similarity determination unit 58, and an actual speed calculation unit 59.

  The speed range setting unit 51 is a processing unit that selectively executes processing according to the speed of the moving object. For example, a speed range of “low speed / medium speed / high speed” can be selected according to the magnitude of the speed, and the time interval of image acquisition is changed in each speed range. This is because if the image acquisition time interval is too long compared to the speed of the moving object, the template image will be framed out of the image to be collated, and the number of moving pixels cannot be obtained. Even if the interval is too short, the images are the same and the number of moving pixels cannot be obtained. Therefore, in order to prevent such a situation, means that can be arbitrarily set by the user according to the speed in consideration of the work content is provided. An operation button for changing the speed range of the speed range setting unit 51 is provided on the input means 7. In this embodiment, three ranges of low speed, medium speed, and high speed are illustrated, but two ranges may be used. Further, the speed range setting unit 51 may recognize the change in the speed of the moving body and automatically change the speed range.

  The still image acquisition unit 52 acquires the reference image and the verification target image from the moving image stored in the storage unit 4 at predetermined time intervals. The time interval for acquiring an image is determined in accordance with the speed range set in the speed range setting unit 51. If the moving speed is fast, the time interval is short to prevent frame-out, and if the moving speed is slow, the time interval is long. It is stipulated that

  The direction encoding processing unit 53 cuts out a template image that is the basis of collation from the reference image and acquires the maximum brightness gradient direction of all the pixels constituting the template image as a code. Further, a search area within a predetermined range is selected from the collation target image, and the maximum brightness gradient direction of all the pixels constituting the search area is acquired as a code. In this way, a value (integer value) obtained by quantizing the gradient direction in which the brightness change in the vicinity of the pixel is maximized is acquired as a code, not the brightness of the pixel itself. This code is called a direction code.

ここでΓは低コントラストな画像を排除するための閾値である。水平方向勾配、垂直方向勾配の絶対値の和が閾値Γより大きい場合には０〜Ｎ−１に符号化され、小さい場合にはＮに符号化される。閾値Γを導入した目的は画素近傍のコントラストの低い画素を排除し、安定して勾配方向を求めるためである。但し、あまり閾値Γの値を大きくすると方向符号の情報が抑制される。Ｃ_ｘｙは量子化幅をΔθ＝π／８とすると図３に示すような０〜１５の値をとる。 When the brightness at the pixel (x, y) of the reference image is I (x, y), the horizontal gradient is ∇I _x = ∂I / ∂x, and the vertical gradient is ∇I _y = ∂I / ∂y. The gradient direction θ _{xy of} brightness is expressed by the following equation (1).
θ _xy = tan ⁻¹ (∇I _x / ∇I _y ) (1)
In this embodiment, the Sobel operator is used to calculate the directional gradient by performing computation on pixels in the vicinity of the target pixel. The direction code C _xy is a value obtained by quantizing the gradient direction θ _xy with a quantization width Δθ, and is represented by the following equation (2).

Here, Γ is a threshold value for eliminating low-contrast images. When the sum of the absolute values of the horizontal gradient and the vertical gradient is larger than the threshold Γ, it is encoded as 0 to N−1, and when it is smaller, it is encoded as N. The purpose of introducing the threshold value Γ is to eliminate pixels with low contrast in the vicinity of the pixels and to obtain the gradient direction stably. However, if the value of the threshold Γ is increased too much, the direction code information is suppressed. C _xy takes a value of 0 to 15 as shown in FIG. 3 when the quantization width is Δθ = π / 8.

  Since the above-described direction code distribution has high uniqueness for each image, it becomes a feature value for representing the image. In addition, the lightness gradient direction is a feature quantity that is robust (invariable) with respect to lightness change because the influence of lightness change is small. Therefore, it can be said that it is a suitable comparison value for obtaining the similarity with respect to a pair of ground images in which there are many defect conditions such as brightness change, shielding, and noise as in an actual field.

  Note that the direction encoding processing unit 53 extracts a plurality of template images from the predetermined position of the reference image, that is, three template images in the present embodiment. This prevents errors from occurring and increases the accuracy of calculating the number of moving pixels.

Next, as shown in FIG. 4, the direction code collation unit 54 collates the template image with each image while sequentially moving the template image in the horizontal direction and the vertical direction. In other words, the sum of absolute values of residuals that compares the sign of the maximum gradient direction of the template image and the sign of the maximum gradient direction of the search area selected from the comparison target image is used, and this is divided by the total number of pixels. The absolute value of the difference is obtained, and the position where this difference is minimum is detected as the collation position. As shown in FIG. 4, the reference image with an image size of M × M is If, the target image is Ig, the codes for each matching pixel are (C _If , C _Ig ), the residual absolute value is d, the average residual is When the difference absolute value is D, it is expressed by the following equation (3).

  In addition, the direction code matching unit 54 performs comparison with the previous residual value in the process of searching for the minimum value in order to speed up the calculation of the direction code matching process, and ends the calculation when it becomes large, A censoring process is introduced to proceed to the next collation calculation. Thereby, in the case of images having low similarity, the number of calculations is greatly reduced.

  Next, as a result of the collation by the direction code collation unit 54, the same similar image specifying unit 55 is the same similarity that most closely approximates an image in which the residual absolute value obtained by comparing the direction code of the template image from the search region is the minimum error. It is specified as an image.

  The moving pixel number calculating unit 56 calculates the difference in the number of pixels between the position of the same similar image specified by the same similar image specifying unit 55 and the position of the initial template image as the moving pixel number. Further, as a result of determining the moving pixel number difference by the moving pixel number difference determining unit 57, which will be described later, if there is a template image in which the moving pixel number difference is equal to or less than a predetermined threshold, the average moving pixel number is calculated from the moving pixel number. Is calculated. Further, although none of the difference in the number of moving pixels is less than or equal to a predetermined threshold, the similarity is determined by the similarity determination unit 58 described later, and as a result, the template image having the similarity less than or equal to the predetermined threshold and its template image When the same similar image exists, the number of moving pixels is specified as the number of moving pixels used for speed calculation.

  The moving pixel number difference discriminating unit 57 is for obtaining a more accurate moving pixel number. The moving pixel number difference determining unit 57 obtains a difference in moving pixel number between a plurality of template images and determines whether or not the difference is equal to or less than a predetermined threshold value. It is to determine whether. In the present embodiment, since the number of moving pixels of the three template images is obtained, the moving pixel number difference determining unit 57 also determines whether or not the difference between the three images is equal to or less than a threshold value. This determination method will be described. Of the three template images, the second template image is cut from the center position of the reference image. For this reason, since it is an image corresponding to the center of the photographic lens, it is considered to be an image with the least distortion and high reliability. Therefore, when comparing the difference in the number of moving pixels, the difference from the other first template image and the difference from the third template image are respectively compared around the second template image. The difference between the third and third template images is not determined as a threshold, and efficiency is improved.

  When the difference in the number of moving pixels in the three template images is not less than or equal to a threshold value, the similarity determination unit 58 compares the template image with the same similar image and the similarity is equal to or less than a predetermined threshold value. Whether or not it exists is determined. This similarity determination may be made by comparing all three template images and determining the template image having the highest similarity, or specifying the template image having a similarity equal to or less than a threshold value. Alternatively, only the central second template image may be compared and determined.

  Here, a moving pixel number difference determining method for the three template images by the moving pixel number difference determining unit 57 and an average pixel number calculating method will be described with reference to FIG. Case 1 is a case where the difference in the number of moving pixels of the three template images is less than or equal to the threshold value. In this case, the moving pixel number calculating unit 56 calculates an average moving pixel number obtained by averaging the three moving pixel numbers. Case 2 and Case 3 are cases where the difference between the two template images is equal to or less than the threshold value. In this case, the average value of the two moving pixels below the threshold, specifically, the number of moving pixels of the second template image is doubled, and the difference in the number of moving pixels below the threshold is adjusted to be ½. The calculated value is calculated as the number of moving pixels. On the other hand, Case 4 is a case where no difference in the number of moving pixels is below the threshold value. In this case, an image pair in which the similarity between the template image and the same similar image is equal to or less than the threshold is obtained, and the number of moving pixels is used as a basis for speed calculation. If none of the similarities is below the threshold, the speed calculation based on the number of moving pixels is not performed.

Subsequently, the actual speed calculation unit 59 calculates the actual speed V of the moving object based on the number of moving pixels calculated by the moving pixel number calculation unit 56 described above. Specifically, based on the speed measurement principle as shown in FIG. 2, the height H of the camera lens from the field, the focal length f of the camera, the acquisition frame interval t, the number of moving pixels m, the size L of the light receiving element, Then, it calculates with the following formula | equation (4).

  Each configuration of the present embodiment has the above configuration. Considering image acquisition and collation here, in the process of repeatedly extracting a template image from one reference image, performing collation with the next collation target image, and extracting the template image at the same time, Due to the relationship, the frame acquisition interval becomes variable, and the number of moving pixels changes for each verification. In order to avoid this, the processing time T2 is set so that the reference image from which the template image is extracted and the target image to be collated are acquired with a certain interval t, and then collation is performed during the collation time T1. Thus, it can be said that the number of moving pixels is not significantly changed, and processing such as limitation of the collation range and limitation of the direction encoding range becomes effective. FIG. 6 shows the status of such processing over time.

  Next, the operation of the speed sensor 1 and the operation of the speed sensor program in the present embodiment will be described with reference to the flowchart of FIG.

  First, when operating the speed sensor 1 of this embodiment, the power switch of the input means 7 is turned on and the speed sensor 1 is started (start). The switch of the photographing means 2 may be linked to this power switch, or a separate photographing start switch may be provided. Subsequently, when the speed range corresponding to the work content is selected by operating the input means 7, the speed range selected by the speed range setting unit 51 is set (step S1). Thereby, the still image acquisition unit 52 acquires a still image as a reference image from the moving image shot by the shooting unit 2, and stores it in a predetermined memory unit in the storage unit 4 (step S2). Subsequently, the direction encoding processing unit 53 acquires three template images from the reference image (step S3). At this time, since the second template image is acquired from the center position of the reference image, there is no distortion and the most accurate reliability is obtained. And the direction encoding process part 53 calculates | requires the maximum brightness gradient direction of each pixel about three template images, and acquires this as a direction code (step S4).

  Next, after a predetermined time has elapsed, the still image acquisition unit 52 extracts a still image that is a verification target image from the moving image and stores it in a predetermined memory unit in the storage unit 4 (step S5). For the collation target image, the direction encoding processing unit 53 sets a search area within a predetermined range, obtains the maximum brightness gradient direction, and acquires it as a direction code (step S6).

  Subsequently, the direction code collation unit 54 collates the direction code of the template image with the direction code in the search area (step S7). In this collation, as shown in FIG. 4, the collation is advanced by sequentially moving the template image in the horizontal direction and the vertical direction in the search area. In this direction code collation, the direction codes of the respective pixels are compared to determine the residual absolute value, the total sum thereof is acquired, and the average residual absolute value obtained by dividing by the total number of pixels is acquired. The image having the smallest average residual absolute value is specified as the same similar image (step S8). In such collation and identification, for the purpose of improving the efficiency and speed of calculation, the comparison is made with the residual value that has already been obtained, and the added value of the residual absolute value is the same as the previous residual absolute value. When it becomes larger than the added value, the further addition processing is discontinued, and the process proceeds to the next collation calculation.

  When the same similar image closest to the template image is acquired by such processing, the difference in the number of pixels moved from the template image to the same similar image is calculated to obtain the number of moving pixels (step S9). Thereby, the number of moving pixels of the first sheet is obtained. Subsequently, in step S10, it is determined whether or not the calculation of the number of moving pixels has been completed for all template images. If other template images remain (in the case of NO), the process returns to step S7, and the next The matching process is repeated for the template image.

  On the other hand, if the number of moving pixels of all template images has been calculated (in the case of YES), the process proceeds to step 11. In step 11, the respective differences are calculated for the number of moving pixels of the three template images. Specifically, if the number of moving pixels of the first template image is d1, the number of moving pixels of the second template image is d2, and the number of moving pixels of the third template image is d3, the difference in the number of moving pixels Are determined as (d2-d1) and (d2-d3). Whether or not to apply to the calculation of the actual speed is finally determined according to the difference in the number of moving pixels.

  First, the moving pixel number difference discriminating unit 57 determines whether or not the moving pixel number difference between the three template images is equal to or less than a predetermined threshold (step S12). If YES, that is, if the difference in the number of three moving pixels is equal to or less than the threshold value, the process proceeds to step 13 and the moving pixel number calculating unit 56 obtains the average value of the moving pixel numbers of the three template images. Based on this average moving pixel number, the actual speed calculation unit 59 calculates the speed (step 18), and the speed display means 6 displays the speed (step S19). On the other hand, if the determination in step S12 is NO, that is, if the difference between the three moving pixel numbers is not less than the threshold value, the process proceeds to step S14.

  In step S14, it is determined whether or not the difference in the number of moving pixels between the two template images is equal to or less than a threshold value. If YES, that is, if either (d2-d1) or (d2-d3) is equal to or smaller than the threshold value, the process proceeds to step S15, and the number of moving pixels of two template images equal to or smaller than the threshold value is determined. The average number of moving pixels is obtained. Based on this average moving pixel number, the actual speed calculation unit 59 calculates the speed (step 18), and the speed display means 6 displays the speed (step S19). On the other hand, if the determination in step S14 is NO, that is, if the difference between the two moving pixel numbers is not less than the threshold value, the process proceeds to step S16.

  In step S16, it is determined whether or not the degree of similarity between each template image and each identical similar image that approximates the template image is equal to or less than a predetermined threshold value. In the case of YES, that is, when the similarity based on the average residual absolute value is equal to or smaller than a predetermined threshold value, the process proceeds to step S17, and the number of moving pixels of the template image is determined as a basis for speed calculation. Based on the number of moving pixels, the actual speed calculation unit 59 calculates the speed (step S18), and displays the speed on the speed display means 6 (step S19). On the other hand, in the case of NO, that is, when there is no identical similar image having a similarity equal to or less than the threshold value for any template image, the process returns to step S2 and starts again from acquiring the reference image.

  When the speed of the speed sensor 1 is calculated by the processing as described above, an agricultural chemical spraying device, a fertilizer spraying device, a crop digging device, etc. (not shown) are connected to the speed sensor 1. It is possible to proceed accurately with the speed.

  According to the present embodiment as described above, it is possible to measure the actual speed of the agricultural tractor and the like in the field without being affected by the slip of the tire, and in conjunction with the speed of the agricultural tractor, pesticide application and fertilizer application, Work can be performed with high precision when harvesting agricultural products.

  In addition, since the speed of the moving object is calculated by applying the direction code matching method based on image processing, it is possible to maintain a certain level of accuracy even under severe conditions such as changes in brightness and shielding, while reducing costs. The speed can be calculated while holding

  Further, in the actual direction code matching process, when searching for the minimum value of the average residual absolute value, the comparison is made with the previous residual value and the calculation is stopped and the next matching calculation is performed. Therefore, the processing load is reduced and high-speed processing is possible.

  The speed sensor 1 and the speed sensor program according to the present invention are not limited to the above-described embodiment, and can be changed as appropriate.

1 is a configuration block diagram showing an embodiment of a speed sensor according to the present invention. It is a schematic diagram which shows the speed measurement principle of this embodiment. It is a schematic diagram explaining the direction encoding process in this embodiment. It is a schematic diagram explaining the direction code collation process in this embodiment. It is a schematic diagram explaining the discrimination | determination process of the moving pixel number difference of the three template images in this embodiment. It is a schematic diagram explaining the acquisition time interval of the still image in this embodiment. It is a flowchart figure which shows the processing flow of the program in this embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Speed sensor 2 Imaging | photography means 3 Illumination means 4 Memory | storage means 5 Calculation processing means 6 Speed display means 7 Input means 51 Speed range setting part 52 Still image acquisition part 53 Direction coding process part 54 Direction code collation part 55 Same similar image specification part 56 moving pixel number calculating unit 57 moving pixel number difference determining unit 58 similarity determining unit 59 actual speed calculating unit

Claims (9)

An imaging unit for imaging the ground from a moving body, a storage unit for storing a captured image captured by the imaging unit, a single still image obtained from the storage unit, a predetermined template image being cut out, and a predetermined time A speed sensor having an arithmetic processing unit that calculates a speed of the moving body by obtaining a still image after elapse and checking the template image to obtain the number of moving pixels,
The arithmetic processing means includes:
A still image acquisition unit that acquires a reference image that is a reference for speed calculation, and that acquires a collation target image that is a collation target after a lapse of a predetermined time;
A template image is cut out from the reference image and the maximum brightness gradient direction of the pixels constituting the template image is obtained as a code, and a search area within a predetermined range is selected from the collation target image to obtain the pixels constituting the search area. A direction encoding processing unit that acquires the maximum brightness gradient direction as a code;
The sum of absolute values of residuals obtained by comparing the sign of the maximum gradient direction of the template image and the sign of the maximum gradient direction of the search region is obtained, and collation is performed using an average residual absolute value obtained by dividing the sum by the total number of pixels. A direction code matching unit to perform,
As a result of the collation, an identical similar image identifying unit that identifies an image having the smallest average residual absolute value as the identical similar image of the template image in the search region;
A moving pixel number calculating unit that calculates the number of moving pixels by comparing the template image and the same similar image;
An actual speed calculator that calculates the actual speed of the moving body based on the calculated number of moving pixels ,
The direction encoding processing unit or the moving pixel number calculating unit cuts out a plurality of template images from one reference image, executes direction encoding processing, and outputs the same similar image that is the same as or similar to each template image. While calculating the number of moving pixels,
Further, the arithmetic processing means obtains a difference of the movement number of pixels in the plurality of sheets of template images, and the number of pixels moved difference determination unit that the difference is determined whether to or less than a predetermined threshold value,
A similarity determination unit that determines whether the similarity based on the direction code between the template image and the same similar image is equal to or less than a predetermined threshold when the difference in the number of moving pixels is not less than the threshold. And
Based on the determination results of these determination units, when the moving pixel number calculation unit determines that the moving pixel number difference is present and there is a template image in which the moving pixel number difference is equal to or less than a predetermined threshold, those moving pixels On the other hand, the average moving pixel number is calculated from the number, and on the other hand, the difference between the moving pixel number is not less than a predetermined threshold, but as a result of the similarity determination, the template image having the similarity lower than the predetermined threshold A speed sensor characterized by calculating the number of moving pixels when the same similar image exists. In claim 1 , when the direction code collating unit compares the direction codes between the pixels to obtain a sum of residual absolute values, the direction code collating unit compares the sum of the previous residual absolute values with a sum equal to or greater than the sum. A speed sensor characterized in that when it becomes larger, the calculation process is stopped at that point and the process proceeds to a collation process with the next image in the search area. In Claim 1 or Claim 2 , the arithmetic processing means has a speed range setting unit that can arbitrarily set a time interval from the acquisition of the reference image to the acquisition of the verification target image according to the speed of the moving body. Features speed sensor. 4. The direction encoding processing unit according to claim 1 , wherein the direction encoding processing unit cuts out three template images from one reference image, and selects a second template image from the center position of the reference image. In the case of cutting out, the moving pixel number difference determination unit performs the difference in the moving pixel number between the second template image and the first template image, and the difference in the moving pixel number between the second template image and the third template image. A speed sensor that calculates whether each difference is equal to or less than a predetermined threshold. A storage unit that stores a photographed image of the ground imaged from the moving body by the imaging unit, and acquires a single still image from the storage unit, cuts out a predetermined template image, and acquires a still image after a predetermined time has elapsed. Then, an arithmetic processing means for calculating the speed of the moving object by obtaining the number of moving pixels by collating with the template image, and a speed sensor program for causing the computer to execute as a display means for displaying the calculated speed,
A reference image acquisition step in which the arithmetic processing means acquires a reference image serving as a reference for speed calculation from the storage means;
A template image direction encoding step of cutting out a template image from the reference image and acquiring the maximum brightness gradient direction of the pixels constituting the template image as a code;
A collation target image obtaining step of obtaining a collation target image at a predetermined time interval after obtaining the reference image;
A search region direction encoding step of selecting a search region within a predetermined range from the collation target image and acquiring the maximum brightness gradient direction of pixels constituting the search region as a code;
The sum of absolute values of residuals obtained by comparing the sign of the maximum gradient direction of the template image and the sign of the maximum gradient direction of the search region is obtained, and collation is performed using an average residual absolute value obtained by dividing the sum by the total number of pixels. A direction code matching step,
As a result of this collation, the same similar image specifying step of specifying an image having the smallest average residual absolute value as the same similar image of the template image in the search area;
A moving pixel number calculating step of calculating the moving pixel number by comparing the template image and the same similar image;
And causing the computer to execute an actual speed calculation step of calculating an actual speed of the moving body based on the calculated number of moving pixels ,
In the template image direction encoding step or the moving pixel number calculating step, a plurality of template images are cut out from one reference image and a direction encoding process is executed, and the same similar image that is the same as or similar to each template image And calculating the number of moving pixels,
Furthermore, a difference between the number of moving pixels in the plurality of template images is obtained, and a moving pixel number difference determining step for determining whether the difference is equal to or less than a predetermined threshold value;
As a result of the determination of the moving pixel number difference, when there is a template image in which the moving pixel number difference is equal to or less than a predetermined threshold, an average moving pixel number calculating step of calculating an average moving pixel number from the moving pixel number;
As a result of the determination of the difference in the number of moving pixels, if none of the difference in the number of moving pixels is less than or equal to a threshold value, whether or not the similarity based on the direction code between the template image and the same similar image is equal to or less than a predetermined threshold value. A similarity determination step to determine;
As a result of the similarity determination, if there is a template image whose similarity is equal to or less than a predetermined threshold and the same similar image, a similarity reference moving pixel number calculating step for calculating a moving pixel number based on the template image A speed sensor program characterized in that it is executed. 6. In the direction code collation step according to claim 5 , when the direction codes between the pixels are compared to determine the sum of the residual absolute values, the sum of the previous residual absolute values is compared, A speed sensor program characterized by stopping the calculation process at that point and shifting to a collation process with the next image in the search area when it becomes large. According to claim 5 or claim 6, speed range setting set according to the speed range of the time interval when the speed range of the moving body is changed, the reference image from the acquisition to acquisition of comparison target image by the speed range setting unit A speed sensor program characterized by causing a step to be executed. The template image direction encoding step according to any one of claims 5 to 7 , wherein three template images are cut out from one reference image, and a second template image is selected as a center position of the reference image. In the moving pixel number difference determining step, the difference in the moving pixel number between the second template image and the first template image, and the moving pixel number difference between the second template image and the third template image are determined. A speed sensor program that calculates differences and determines whether each difference is equal to or less than a predetermined threshold.   A moving body comprising the speed sensor according to any one of claims 1 to 4 or the speed sensor program according to any one of claims 5 to 8.

Images