JP6300571B2 - Harvest assistance device - Google Patents

Description

  The present invention relates to a harvesting assistance device. In particular, the present invention relates to a harvesting assistance device that can accurately detect a harvesting target to be harvested.

  Conventionally, various harvesting apparatuses have been proposed that capture an image of a harvested object such as a vegetable or fruit and perform image processing on the captured image to detect the harvested object. In general, this type of harvesting apparatus captures a grayscale image composed of pixels having density values according to the amount of reflected light reflected from a harvested object by an imaging unit, and based on the grayscale difference between the pixels of the captured grayscale image. This is a method for detecting an object to be harvested.

  As a harvesting device as described above, for example, Patent Document 1 discloses that a grayscale image captured by a CCD camera (a grayscale image obtained by performing a predetermined process on an RGB image) is binarized, and the binarized image is used. There has been proposed a harvesting device that estimates the length (fruit length) of a harvesting object and detects the harvesting object to be harvested based on the estimated length. That is, in the harvesting device described in Patent Document 1, the harvesting target to be harvested is detected based on the grayscale image corresponding to the amount of reflected light from the harvesting target (paragraph 0016 and the like in Patent Document 1). In addition, although the ultrasonic distance sensor for measuring the distance to a harvest target object is disclosed by patent document 1, this ultrasonic distance sensor is used in order to detect the harvest target object which should be harvested. Instead, it is used exclusively to bring a manipulator equipped with a CCD camera and cutting means (fruit scissors) closer to the harvested object (paragraph 0015, etc. of Patent Document 1).

  An apparatus for detecting a harvested object to be harvested based on a grayscale image, such as the apparatus described in Patent Document 1, can be said to be an effective means when the grayscale difference between the harvested object and the background is large. However, it is difficult to detect the harvested object separately from the background if the harvested object and the background are close to each other, such as when the harvested object grows densely. Therefore, it is difficult to detect the harvest object to be harvested.

Japanese Patent No. 3757279

  The present invention has been made in order to solve such problems of the prior art, and an object of the present invention is to provide a harvesting assisting device that can accurately detect a harvesting target to be harvested.

  In order to solve the above-described problem, the present invention includes a two-dimensional coordinate of a harvesting object on a plane orthogonal to a growing direction of the harvesting object and a harvesting object along the growing direction in the two-dimensional coordinate of the harvesting object. A three-dimensional position measuring means for measuring the distance between the two and a signal processing means for determining a harvesting target to be harvested, wherein the signal processing means is measured by the three-dimensional position measuring means. A distance image forming means for forming a distance image composed of pixels having two-dimensional coordinates and a density value corresponding to the distance measured by the three-dimensional position measuring means, and the distance image has a predetermined distance or less. Edge detection means for detecting an edge of a pixel area composed of the pixels having a corresponding density value, and the detected edge is determined in advance according to a harvest object to be harvested. And determining means for determining that the pixel located at the two-dimensional coordinates of the matched geometric shape is the edge of the harvested object to be harvested in the distance image by pattern matching. Provided is a harvesting assisting device.

  According to the harvesting assisting device of the present invention, the three-dimensional position measuring means moves along the two-dimensional coordinates of the harvesting object in a plane orthogonal to the growing direction of the harvesting object and the growing direction in the two-dimensional coordinates of the harvesting object. The distance to the harvested object is measured. Also, a distance image formed by pixels having two-dimensional coordinates measured by the three-dimensional position measuring means and a density value corresponding to the distance measured by the three-dimensional position measuring means by the distance image forming means of the signal processing means. Is formed. For this reason, the two-dimensional coordinates of the distance image correspond to the two-dimensional coordinates of the harvest object, and the density values of the pixels constituting the distance image correspond to the distance to the harvest object in the two-dimensional coordinates where the pixels are located. Will do. The edge detection means of the signal processing means detects an edge of a pixel area composed of pixels having density values corresponding to a predetermined distance or less in the distance image. Since the measured distance to the harvested object (corresponding to the density value in the distance image) has a negative correlation with the distance that the harvested object extends in the growth direction, the density corresponding to a predetermined distance or less in the distance image By detecting the edge of the pixel region composed of pixels having a value, only the edge of the harvesting object extending beyond a predetermined height in the growth direction can be selected. A distance image composed of pixels having a density value corresponding to the distance is different from a grayscale image composed of pixels having a density value corresponding to the amount of reflected light as in the conventional case, and is affected by the amount of reflected light of the background of the harvest target Not receive. For this reason, since the distance from the three-dimensional position measurement unit differs between the harvest object and the background, the pixel area corresponding to the harvest object and the pixel area corresponding to the background of the harvest object are obtained by using the distance image. Can be separated with high accuracy. Furthermore, it is possible to accurately select only the harvesting object extending beyond the predetermined height in the growing direction from the harvesting object without using a distance measuring means different from the three-dimensional position measuring means. it can.

  The determination means of the signal processing means pattern-matches a predetermined geometric shape with respect to the detected edge in accordance with a harvest object to be harvested (a harvest object that has grown to be suitable for harvesting). In the distance image, it is determined that the pixel located at the two-dimensional coordinate of the matched geometric shape is the edge of the harvested object to be harvested. By pattern matching a predetermined geometric shape according to the harvested object to be harvested with respect to the detected edge, a harvested object having a desired shape and size in a plane orthogonal to the growth direction can be obtained. It can be a detection target. That is, a predetermined geometric shape that matches only one of the selected edges after selecting only the edge of the harvesting object extending in the growth direction by a predetermined height or more by the edge detection unit as described above. (In the distance image, the pixel located at the two-dimensional coordinates of the matching geometric shape) is determined (imitation) as the edge of the harvested object to be finally harvested. For this reason, whether or not to match the desired shape and size in the plane orthogonal to the growth direction, compared to the case where it is determined whether or not to harvest based on whether or not it extends beyond the predetermined height in the growth direction. Therefore, it can be expected to accurately detect the harvest object to be harvested. For example, even if the edges of a plurality of harvesting objects detected by the edge detection means overlap and appear to have one edge that has an apparently too large dimension (thus considered not to be harvested), By pattern matching predetermined geometric shapes, the geometric shapes that match the actual edges of the individual harvested objects are detected and correctly determined to be the edges of the harvested objects to be harvested I can expect that. Conversely, even if the edges overlap, the result is a pattern with a predetermined geometric shape, even if the edges appear to have an appropriate dimension (and therefore should be harvested). By matching, it can be expected that a geometric shape that matches the actual edge of each harvested object is detected and correctly determined that it is not the edge of the harvested object to be harvested.

  Further, according to the harvesting assisting device according to the present invention, it is possible to separate the harvesting target to be harvested using the same distance image obtained by the one three-dimensional position measuring means, and at the same time, the harvesting target to be harvested. The distance to the object can also be calculated. For this reason, a distance sensor such as an ultrasonic distance sensor for measuring the distance separately from a CCD camera for obtaining a grayscale image of the harvested object in order to bring the cutting means closer to the harvested object as in the prior art. Therefore, the apparatus configuration can be simplified.

  Preferably, the determination means pattern-matches a plurality of circular or elliptical shapes having different diameters as the predetermined geometric shape with respect to the detected edge, and the matched geometric shape is obtained. Categorizing into geometric shape groups that are aggregates of geometric shapes whose center positions are close to each other, and for each geometric shape group, the geometric shapes constituting the geometric shape group Among them, a geometric shape having the highest degree of matching is extracted, and a geometric shape having a predetermined diameter is selected from the geometric shapes extracted for each of the geometric shape groups, In the distance image, it is determined that the pixel located at the two-dimensional coordinate of the selected geometric shape is the edge of the harvested object to be harvested.

  According to such a preferable configuration, the determination unit of the signal processing unit pattern-matches a plurality of circular or elliptical shapes having different diameters as a predetermined geometric shape with respect to the detected edge. By pattern-matching a circular shape or an elliptical shape as a predetermined geometric shape, a plane orthogonal to the growth direction of the harvested object measured by the three-dimensional position measuring means (seeing the harvested object from the growth direction). The possibility of matching with a harvested object having a circular or elliptical shape (for example, mushrooms, etc.) increases. Further, by using a plurality of circular or elliptical shapes having different diameters, the possibility of matching geometric shapes having diameters closer to the actual edges of the harvested object increases.

  However, the number of matching geometric shapes is not necessarily one for an actual edge of one harvesting target, and a plurality of geometric shapes having different diameters may be matched. For example, it is conceivable that a geometric shape having a diameter to be harvested matches a geometric shape having a diameter not to be harvested with respect to an actual edge of a harvested object that should not be harvested. . Therefore, according to the preferred configuration, the matched geometric shapes are classified into a geometric shape group that is an aggregate of geometric shapes whose central positions are close to each other. When a plurality of geometric shapes match an actual edge of one harvested object, the positions of the centers of the matched geometric shapes are considered to be close to each other. Therefore, by classifying into geometric shape groups that are aggregates of such geometric shapes, the matched geometric shapes are classified as geometric shape groups for each actual edge of the harvested object. Can be expected. And according to the said preferable structure, the geometric shape with the highest matching degree is extracted among the geometric shapes which comprise a geometric shape group for every geometric shape group. Of the geometric shapes that make up the geometric shape group (geometric shapes that can be expected to match the actual edge of one harvested object), the geometric shape with the highest degree of matching is extracted. Therefore, it can be expected to extract a geometric shape having a diameter closer to the actual edge of the harvested object. And according to the said preferable structure, the geometric shape which has a predetermined diameter (equivalent to the diameter of the harvesting object mentioned above) out of the geometric shapes extracted for every geometric shape group. A pixel that is selected and is located in the two-dimensional coordinates of the selected geometric shape in the distance image is determined to be the edge of the harvested object to be harvested. As described above, the geometric shape extracted for each geometric shape group can be expected to be a geometric shape having a diameter closer to the actual edge of the harvested object. For this reason, by selecting a geometric shape having a harvest target diameter from the geometric shapes extracted for each geometric shape group, a circular shape having a harvest target diameter selected after pattern matching or It can be expected that the edge of the harvested object is to be harvested with higher accuracy than in the case where only the elliptical shape is subjected to pattern matching.

  Specifically, for example, when the size of the harvested object to be harvested is larger than the shape corresponding to the circular shape having a diameter of 40 pixels, the harvesting target corresponding to the circular shape having a diameter of 37 pixels is harvested. It is desirable to determine that it is not a harvest object to be harvested. If only a circular shape with a diameter of 40 pixels is pattern-matched to an edge detected from a harvested object corresponding to a circular shape with a diameter of 37 pixels, the circular shape with a diameter of 40 pixels matches and the harvesting is performed. There is a risk that it is determined that the edge of the harvested object is to be harvested. On the other hand, when a plurality of circular shapes of a circular shape having a diameter of 35 pixels and a circular shape having a size of 40 pixels are subjected to pattern matching with respect to an edge detected from a harvested object corresponding to a circular shape having a diameter of 37 pixels. In some cases, a circular shape having a diameter of 35 pixels and a circular shape having a size of 40 pixels are matched. In this case, it is considered that the positions of the centers of both circular shapes are close to each other. Both circular shapes are classified into one geometric shape group. For one geometric shape group, it is considered that the circular shape having a diameter of 35 pixels has a higher matching degree than the circular shape having a diameter of 40 pixels. As a result, by extracting a circular shape having a diameter of 35 pixels, it can be expected to more accurately detect the shape of the harvested object to be harvested. For this reason, it can be expected that the extracted geometric shape is not selected as a geometric shape having a diameter to be harvested. That is, it can be expected that the edge is not the edge of the harvested object to be harvested. Thus, it is more accurate to pattern match a plurality of circular shapes of a 35 pixel diameter and a 40 pixel circular shape than to pattern match only a circular shape having a diameter of 40 pixels. It can be expected to determine what to harvest.

  Here, in the present invention, the geometric shapes whose center positions are close to each other can be exemplified as any distance between the centers of the geometric shapes being within a predetermined range. In the present invention, the matching geometric shape is classified into the geometric shape group, in addition to classifying the matched geometric shape into a plurality of geometric shape groups. It is a concept including classification into shape groups. In this case, the centers of all matched geometric shapes are close to each other. In the present invention, the geometric shape group which is an aggregate of geometric shapes includes not only a plurality of geometric shapes but also a single geometric shape. It is a concept. In this case, one geometric shape constituting the geometric shape group is extracted as a geometric shape having the highest degree of matching.

  Preferably, the image processing device further includes a cutting unit for cutting the harvesting object, wherein the signal processing unit 2 of the pixels constituting the pixel region having the edge of the harvesting object determined to be harvested by the determination unit. The apparatus further comprises cutting position determining means for determining a cutting position of the harvested object determined to be harvested based on the dimensional coordinates and the density value, and the cutting means is the cutting determined by the cutting position determining means. It moves to a position and cuts the harvested object determined to be harvested.

  According to such a preferable configuration, the edge of the harvested object that has been determined to be harvested by the cutting position determining unit of the signal processing unit (the pixel located at the two-dimensional coordinate of the matched geometric shape in the distance image) The cutting position of the harvested object determined to be harvested is determined based on the two-dimensional coordinates and the density value of the pixels constituting the pixel area having (). As described above, according to the present invention, the harvest target to be harvested can be detected with high precision (the pixel region having the edge of the harvest target to be harvested can be detected with high precision). The cutting position of the object can also be determined with high accuracy.

  Preferably, the cutting position determining means extracts a pixel located at the center of the edge of the harvested object determined to be harvested in the distance image, and corresponds to the two-dimensional coordinates and density value of the extracted pixel. A three-dimensional position further separated from the three-dimensional position measuring means by a certain distance along the growth direction from the three-dimensional position to be determined is determined as the cutting position.

According to such a preferable configuration, the cutting position determining means is located at the center of the edge of the harvested object determined to be harvested in the distance image (the pixel located at the two-dimensional coordinates of the matched geometric shape). Pixels to be extracted are extracted. That is, the pixel located at the center of the matched geometric shape is extracted. For example, when a circular shape or an elliptical shape is used as the geometric shape, a pixel located at the center of the matched circular or elliptical shape is extracted. The center of the geometric shape is naturally determined according to the geometric shape. That is, if the two-dimensional coordinates of the geometric shape are determined, the two-dimensional coordinates at the center of the geometric shape are naturally determined. Since the pixel located at the center of the matched geometric shape is determined by only the two-dimensional coordinates of the matched geometric shape regardless of the density value of the pixels constituting the distance image, the matched geometric shape is determined. The pixel located at the center of the shape can be easily determined.
In addition, the cutting position determining means sets a three-dimensional position further separated from the three-dimensional position measuring means by a certain distance along the growth direction from the three-dimensional position corresponding to the extracted two-dimensional coordinates and density value of the pixel. It is determined. By moving the cutting means to the determined cutting position and cutting the harvested object determined to be harvested, the length of the harvested object to be cut can be made uniform. Further, as described above, according to the present invention, since the pixel located at the center of the matched geometric shape can be easily determined, the cutting position determined based on this can also be easily determined. .

  The cutting position determination unit is not limited to the preferable configuration, and the cutting position determination unit is configured to calculate the average value and the density value of the two-dimensional coordinates of the pixels constituting the pixel region having the edge of the harvested object that is determined to be harvested by the determination unit. The average value is calculated, and the three-dimensional position further spaced apart from the three-dimensional position measuring means along the growth direction from the three-dimensional position corresponding to the average value of the two-dimensional coordinates and density value of the calculated pixel. May be determined as the cutting position.

  As described above, according to the harvesting assistance device according to the present invention, it is possible to accurately detect a harvesting target to be harvested.

FIG. 1 is a schematic diagram showing a schematic configuration of a harvesting assistance device according to an embodiment of the present invention. FIG. 2 shows an example of the two-dimensional coordinates of the harvested object or medium measured by the three-dimensional position measuring means. FIG. 3 shows an example of an elliptical shape for pattern matching. FIG. 4 is an explanatory diagram for explaining the degree of coincidence when a circular shape is pattern-matched to a detected edge. FIG. 5 is an explanatory diagram for explaining the overlap to be evaluated. FIG. 6 shows an example of a circular shape matched to the detected edge. FIG. 7 shows the result when the edge of the mushroom is determined using the gray image of the mushroom. FIG. 8 shows a result when a mushroom edge is determined using a mushroom distance image.

  Hereinafter, a harvesting assistance device according to an embodiment of the present invention will be described with reference to the accompanying drawings, taking as an example a case where a harvesting target is a mushroom. FIG. 1 is a schematic diagram showing a schematic configuration of a harvesting assistance device 100 according to an embodiment of the present invention. As shown in FIG. 1, the harvest assistance device 100 according to the present embodiment includes a three-dimensional position measurement unit 1 and a signal processing unit 2. Moreover, the harvest assistance apparatus 100 according to the present embodiment further includes a cutting means 3 for cutting the harvest target K.

  FIG. 1 is a schematic view seen from a direction perpendicular to the growth direction of the harvested object K. FIG. As shown in FIG. 1, the harvest object K of the present embodiment grows upward from a medium B called a fungus bed. The growth direction of the harvested object K of the present embodiment is the vertical direction, specifically the Z direction shown in FIG. Moreover, the plane orthogonal to the growth direction of the harvested object K of the present embodiment is a horizontal plane, specifically, an XY plane defined by the X direction and the Y direction shown in FIG.

  In the present embodiment, the harvest object K (mushroom group) grows from the medium B in a randomly arranged state on the XY plane shown in FIG. The three-dimensional position measuring means 1 includes a two-dimensional coordinate of the harvest target K on a plane (XY plane) orthogonal to the growth direction of the harvest target K, and the harvest target K along the growth direction in the two-dimensional coordinate. Measure distance. The signal processing means 2 determines the harvesting target K to be harvested based on the two-dimensional coordinates of the harvesting target K measured by the three-dimensional position measuring means 1 and the distance to the harvesting target K. Then, the cutting means 3 cuts the harvest target K to be harvested determined by the signal processing means 2. In the present embodiment, the three-dimensional position measuring unit 1 measures the two-dimensional coordinates of the harvesting target K and the distance to the harvesting target K, the signal processing unit 2 determines the harvesting target K to be harvested, and cuts it. A series of steps of cutting the harvesting target K to be harvested by the means 3 is executed for each measurement unit region. That is, the three-dimensional position measuring unit 1 is perpendicular to the growth direction of the harvesting target K so that the two-dimensional coordinates of the harvesting target K and the distance to the harvesting target K can be measured for each measurement unit region. It can move in a direction (for example, the X direction and the Y direction shown in FIG. 1).

Here, the measurement unit region is the growth direction of the harvested object K (mushroom group) without moving the three-dimensional position measuring means 1 at the position (three-dimensional position) where the three-dimensional position measuring means 1 is arranged. Means a measurable region on a plane (XY plane) orthogonal to W ₁ and W _{2 in} FIG. 1 indicate measurement unit areas. W _n means an n-th measurement unit area. In this embodiment, the measurement unit area is set to a rectangular shape. Further, as shown in FIG. 1, the measurement unit area W ₁ is adjacent to the measurement unit area W ₂ and X-direction, are partially overlapping. In addition, the measurement unit region W ₃ (not shown) is adjacent to the measurement unit region W ₂ in the Y direction and partially overlaps. The dimensions of each measurement unit area (the dimension in the X direction and the dimension in the Y direction) are set when the center of the harvested object K (mushroom umbrella) to be harvested and the center of the measurement unit area coincide with each other on the XY plane. , At least the harvested object K (mushroom umbrella) having the largest diameter to be harvested is set.

Specifically described above series of steps, in this embodiment, first, the crop object K which grow in the measurement unit area W _1, 3-dimensional position measuring means 1 and the two-dimensional coordinates of the crop object K The distance to the harvesting target K is measured, the signal processing means 2 determines the harvesting target K to be harvested, and the cutting means 3 cuts the determined harvesting target K to be harvested. Then, three-dimensional position measuring means 1, two-dimensional coordinates of crop object crop objects are growing in different units of measurement area W ₂ is a measurement unit area W ₁ which was cut in K K (Mushrooms) And the distance to the harvested object K is moved in the X direction so that it can be measured. The three-dimensional position measuring means 1 indicated by a broken line in FIG. 1 illustrates a state in which the three-dimensional position measuring means 1 indicated by a solid line has moved in the X direction. Next, the crop object K which grow in the measurement unit area W _2, 3-D position measuring means 1 measures the distance to the two-dimensional coordinates and harvesting object K crop object K, the signal processing means 2 The harvesting target K to be harvested is determined, and the cutting means 3 cuts the determined harvesting target K to be harvested. Then, the three-dimensional position measuring unit 1 is configured to collect the harvest target K (mushroom group) growing in a measurement unit region W ₃ (not shown) different from the measurement unit region W _{2 where} the harvest target K is cut. ) And the distance to the harvesting target K so that it can be measured. By repeating the above processing for each harvesting target K (mushroom group) growing in each measurement unit region, the harvesting target K to be harvested is determined for all harvesting targets K and cut. Will be. Details will be described below.

As described above, the three-dimensional position measuring unit 1 uses the two-dimensional coordinates of the harvesting target K on the plane (XY plane) orthogonal to the growth direction of the harvesting target K and the two-dimensional coordinates for each measurement unit region. The distance to the harvesting target K along the growth direction is measured. In FIG. 1, the three-dimensional position measuring means 1 includes five harvesting objects K (K1 to K5) or two-dimensional coordinates of the medium B and the harvesting object K (K1 to K5) or the medium B in the two-dimensional coordinates. The case of measuring the distance (z _{1 to} z ₇ ) is illustrated. FIG. 2 shows an example of the two-dimensional coordinates of the harvest object K (K1 to K5) or the culture medium B measured by the three-dimensional position measuring means 1 shown in FIG. a is a natural number, and (x _a , y _a ) shown in FIG. 2 is an example of the two-dimensional coordinates of the harvested object K or the culture medium B in a plane orthogonal to the growth direction of the harvested object K, as shown in FIG. z _a is an example of the distance to the harvested object K or the medium B along the growth direction in the two-dimensional coordinates (x _a , y _a ). The region _W 1, _{W 2} shown in FIG. 2 shows a three-dimensional position measuring means measuring unit region _W 1 measured by 1, _{W 2} shown in FIG. The three-dimensional position measuring unit 1 includes, for example, an irradiating unit that irradiates the harvested object K with linear light, and an imaging unit that images the irradiated linear light from a direction different from the irradiation direction. The two-dimensional coordinates of the harvesting target K on a plane orthogonal to the growth direction of the harvesting target K and the harvesting target K along the growth direction in the two-dimensional coordinates are obtained by optical cutting using the captured image. A three-dimensional distance meter that measures the distance between the two can be exemplified.

  As described above, the signal processing unit 2 uses the two-dimensional coordinates of the harvesting target K measured by the three-dimensional position measuring unit 1 and the distance to the harvesting target K measured by the three-dimensional position measuring unit 1. Based on this, the harvesting object K to be harvested is determined. Specifically, the signal processing unit 2 of the present embodiment includes a distance image forming unit 21, an edge detection unit 22, and a determination unit 23.

The distance image forming unit 21 forms a distance image composed of pixels having two-dimensional coordinates measured by the three-dimensional position measuring unit 1 and density values corresponding to the distances measured by the three-dimensional position measuring unit 1. To do. Further, the distance image forming unit 21 forms a distance image for each measurement unit region. Specifically, the distance image forming unit 21 forms a distance image in the measurement unit region W ₁ (a distance image shown in the measurement unit region W ₁ in FIG. 2), and a distance image (in the measurement unit region W ₂ ( so on to form a distance image) shown in the measurement unit area W ₂ in FIG. 2, to form a distance image at each measurement unit areas. As described above, the distance image includes pixels having two-dimensional coordinates measured by the three-dimensional position measuring unit 1 and density values corresponding to the distance measured by the three-dimensional position measuring unit 1. For this reason, the two-dimensional coordinates of the distance image correspond to the two-dimensional coordinates of the harvest object K or the culture medium B, and the density value of the pixel constituting the distance image is the harvest object K in the two-dimensional coordinate where the pixel is located. Or it corresponds to the distance to the medium B. In the present embodiment, if the distance measured by the three-dimensional position measuring unit 1 becomes long, the density value of the pixel of the corresponding distance image becomes low. Specifically, the shorter the distance to the harvesting object K or the medium B in the two-dimensional coordinates, the higher the density value of the pixel of the corresponding distance image, to the harvesting object K or the medium B in the two-dimensional coordinates. The longer the distance, the lower the density value of the corresponding distance image pixel. For example, when the density value of the pixels constituting the distance image is represented by 8 bits, the corresponding distance image pixel is 0 (in accordance with the distance to the harvest target K or the culture medium B in the two-dimensional coordinates. That is, it has a density value of black pixels) to 255 (that is, white pixels). Since the three-dimensional position measuring means 1 measures the distance to the harvest target K along the growth direction of the harvest target K, the longer the length that the harvest target K extends in the growth direction, the more 3 The distance measured by the dimension position measuring means 1 is shortened. As shown in FIG. 1, among the harvesting objects K1 to K5, the harvesting object K3 has the longest length in the growth direction. Therefore, in the distance image formed by the distance image forming unit 21, as shown in FIG. Moreover, the density value of the pixel of the distance image corresponding to the two-dimensional coordinates of the harvested object K3 is the highest. The distance image shown in FIG. 2 corresponds to the two-dimensional coordinates of each of the harvest objects K1 to K5 and the culture medium B in order to facilitate understanding of the distance image composed of pixels having density values corresponding to the distance. The density values of the pixels of the distance image to be made are the same.

As described above, in the present embodiment, the signal processing unit 2 determines the harvest object K to be harvested for each measurement unit region. Here, as described above, the harvest object K (mushroom group) grows from the medium B in a randomly arranged state on the XY plane, as shown in FIG. Each measurement unit region corresponds to a part of the culture medium B. For this reason, not only when the harvest target K to be harvested fits in the measurement unit region, but also when a part of the harvest target K to be harvested deviates from the measurement unit region (for example, the harvest target K3 shown in FIG. 2). Can be considered.
In the present embodiment, as shown in FIG. 1, the measurement unit area W ₁ overlaps with the measurement unit area W _2. Specifically, the measurement unit area W ₁ is sized to overlap with the measurement unit area W ₂ is a more maximum dimensions of the crop object K to be harvested. More specifically, when the harvesting target K to be harvested has a circular shape in the measurement unit region, the value is equal to or greater than the maximum value of the diameter of the harvesting target K to be harvested. Further, when the harvesting target K to be harvested has an elliptical shape in the measurement unit region, it is set to be equal to or greater than the maximum value of the major axis of the harvesting target K to be harvested. By setting the overlapping dimension to be equal to or larger than the maximum value of the dimension of the harvesting target K to be harvested, the harvesting target K to be harvested will be contained in any measurement unit region, so that the signal processing means 2 harvests. It can be expected to determine that it is a harvesting target K.
Dimension measuring unit regions W ₁ overlaps with the measurement unit area W ₂ has a case has been described is greater than or equal to the maximum value of the dimension of the crop object K to be harvested, the maximum value of the dimension of the crop object K to be harvested It may be a small value. When the overlapping dimension is smaller than the maximum value of the dimension of the harvested object K to be harvested, it is conceivable that a part of the harvested object K to be harvested deviates from the measurement unit area in any measurement unit area. Even in this case, the signal processing means 2 performs pattern matching to be described later (specifically, a matching score is calculated, and whether the calculated matching score is a predetermined threshold value or more is evaluated. Therefore, it can be expected that it is the harvesting object K to be harvested.
In the present embodiment, although the measurement unit area W ₁ is present measurement unit area W ₂ and overlap portions, the present invention is not limited thereto, overlapping measurement unit area W ₁ is a measurement unit area W ₂ The portion may not exist.

  The edge detection means 22 detects an edge of a pixel area composed of pixels having density values corresponding to a predetermined distance or less in the distance image. In the present embodiment, the edge detection unit 22 forms a binarized image from a distance image using a density value corresponding to a predetermined distance as a threshold value. Specifically, it indicates that the distance from the three-dimensional position measuring unit 1 is a white pixel (for example, a pixel having a density value of 255) indicating that the distance is equal to or less than a predetermined distance, and is greater than the predetermined distance. A binary image composed of black pixels (for example, pixels having a density value of 0) is formed by the edge forming means 22. The reason why the distance from the three-dimensional position measuring means 1 is equal to or less than the predetermined distance is that the harvest object K extends beyond a predetermined height in the growth direction. It is possible to select only the harvesting object K extending from the height to the growth direction to a predetermined height or more with high accuracy. In the present embodiment, the binarization is composed of white pixels indicating that the distance from the three-dimensional position measuring means 1 is equal to or less than a predetermined distance and black pixels indicating that the distance is greater than the predetermined distance. Although the image is formed, the present invention is not limited to this, and the black pixel indicating that the distance from the three-dimensional position measuring means 1 is equal to or smaller than the predetermined distance and the predetermined distance is larger. A binarized image composed of white pixels indicating this may be formed.

  Moreover, the edge detection means 22 detects the edge of the pixel area comprised from the pixel corresponded below a predetermined distance in the binarized image formed from the distance image. In order to detect an edge of a pixel region composed of pixels corresponding to a predetermined distance or less, for example, use of a Sobel filter can be exemplified. Specifically, it is possible to exemplify detecting a change in shading between a pixel constituting the binarized image and a pixel adjacent to the pixel. In the present embodiment, the edge detection unit 22 detects edges in the binarized image, but the present invention is not limited to this, and may detect edges in the distance image. Specifically, it is possible to exemplify detection of a pixel in which a change in shading between a pixel constituting the distance image and a pixel adjacent to the pixel is greater than or equal to a predetermined value.

  The determination unit 23 pattern-matches the edge detected by the edge detection unit 22 with a geometric shape predetermined according to the harvesting target K to be harvested. Specifically, it is detected in the binarized image formed from the distance image by sequentially performing pattern matching with a predetermined geometric shape at a predetermined portion of the detected edge for every predetermined portion. It is detected whether or not there is a portion similar to a predetermined geometric shape in the edge. In the present embodiment, the harvest object K to be harvested is a mushroom having an umbrella having a predetermined size. If you look at the mushroom umbrella from the growth direction, it becomes a shape that approximates a circular shape or an elliptical shape, so the geometric shape to be pattern-matched should be circular, elliptical, or both circular and elliptical become. Hereinafter, the pattern matching process will be specifically described.

  In the present embodiment, a plurality of circular shapes and elliptical shapes having different diameters are used as geometric shapes for pattern matching. Specifically, the ratio between the minor axis and the major axis of the elliptical shape is 0.8. More specifically, in a binarized image formed from a distance image, each circular shape with a diameter of 20 to 80 pixels, a major axis of 20 to 80 pixels, and a minor axis of 16 to 64 pixels. Each elliptical shape up to the pixel is used. Further, as shown in FIG. 3, for each elliptical shape, an angle formed by the X-axis direction (X direction shown in FIG. 3) of the binarized image formed from the distance image and the major axis direction of the elliptical shape (hereinafter referred to as “the elliptical shape”). , The rotation angle) is θ, and each elliptical shape in which the range of 0 ° ≦ θ <180 ° is changed at a pitch of 1 ° is used. In the present embodiment, the circular shape and the elliptical shape having the diameter as described above are used as the geometric shape, but the present invention is not limited to this, and the size of the harvested object K to be harvested is limited. Accordingly, the circular and elliptical diameters to be pattern-matched and the ratio of the elliptical minor axis to the major axis can be changed. In this embodiment, the rotation angle of the elliptical shape is changed at a 1 ° pitch, but the present invention is not limited to this.

The determination unit 23 evaluates the degree of coincidence with the edge detected by the edge detection unit 22 using the plurality of circular shapes and elliptical shapes having different diameters described above. FIG. 4A is an explanatory diagram for explaining the degree of coincidence when the circular shape is pattern-matched to the detected edge, and FIG. 4B is an enlarged view of the region S shown in FIG. It is. Pixels G1 to G7 shown in FIG. 4B are pixels that form a circular pattern to be pattern-matched, and pixels E1 to E10 are pixels that form a detected edge. The degree of coincidence is calculated from the distance between the pixels forming the circular pattern to be pattern-matched and the pixels forming the detected edge. Specifically, the determination unit 23 detects the detected edge on a straight line connecting the center of the pixel G1 and the center of the pixel (pixel C shown in FIG. 4A) positioned at the center of the circular pattern to be pattern-matched. Are extracted. The determination unit 23 calculates a distance between the center of the extracted pixel and the center of the pixel G1, acquires the calculated distance as a positional deviation amount from the pixel G1, and stores it. When there are a plurality of extracted pixels, the shortest distance among the calculated distances is the amount of positional deviation from the pixel G1 (the length of the portion sandwiched between the arrows shown in FIG. 4 (a), Specifically, it is acquired and stored as d ₁ ) shown in FIG. For all the pixels of the remaining composing the circular shape to the pattern matching, and obtains the positional shift amount d ₂ to d _n from the pixel G2~Gn as previously described (n is the number of pixels constituting the circular shape to the pattern matching) ,Remember. Judging means 23, the sum of the positional displacement amount d ₁ to d _n from the stored pixel G1 to Gn, normalized by the number of pixels n constituting the circular shape to the pattern matching, and calculates the matching degree. Even if the geometric shape to be pattern-matched is an elliptical shape, the determination unit 23 can calculate the degree of coincidence in the same manner. By evaluating the calculated degree of coincidence, it is possible to uniformly evaluate circular shapes and elliptical shapes having different diameters and rotation angles. The determination unit 23 stores, as matching candidates, a circular shape or an elliptical shape that is evaluated to have a degree of matching equal to or higher than a predetermined value in the binarized image formed from the distance image.

  Next, the determination means 23 is the same among the matching candidates (in the case of a circular shape, a circular shape having the same diameter, and in the case of an elliptical shape, elliptical shapes having the same major axis, minor axis, and rotation angle). Evaluate whether or not. FIG. 5 is an explanatory diagram for explaining the overlap to be evaluated. Specifically, as shown in FIG. 5, the amount of deviation in the X direction and the Y direction is evaluated for the same shapes in the matching candidates. For example, in the case of an elliptical shape, the amount of deviation in the X direction is the amount of X direction displacement between the end point in the X direction of the pixels constituting one elliptical shape and the end point in the X direction of the pixels constituting the other elliptical shape. It is calculated as the number of pixels between the end points (length in the X direction). Specifically, the absolute value of the difference between the largest X coordinate value among the X coordinates of the pixels constituting one elliptical shape and the largest X coordinate value among the X coordinates of the pixels constituting the other elliptical shape. Is calculated as The amount of deviation in the Y direction is calculated similarly. The determination unit 23 calculates the number of pixels (size in the X direction) between both end points in the X direction of the pixels constituting one elliptical shape. When the calculated amount of deviation in the X direction is 10% or more of the calculated size in the X direction, the determination unit 23 evaluates that both ellipse shapes do not overlap, and both ellipse shapes remain matching candidates. leave. For example, when the size in the X direction is 40 pixels and the amount of deviation in the X direction is 4 pixels or more, it is evaluated that they do not overlap. The size in the Y direction is calculated in the same manner, and when the calculated amount of deviation in the Y direction is 10% or more of the calculated size in the Y direction, the determination unit 23 evaluates that both elliptical shapes do not overlap. Then, both elliptical shapes are left as matching candidates. On the other hand, when the calculated displacement amount in the X direction is less than 10% of the calculated size in the X direction, and the calculated displacement amount in the Y direction is less than 10% of the calculated size in the Y direction. The determination means 23 evaluates that the two elliptical shapes overlap each other. When it is evaluated that the elliptical shapes overlap, the determination unit 23 compares the degree of coincidence of one elliptical shape with the degree of coincidence of the other elliptical shape, and selects the elliptical shape having the higher degree of coincidence as a matching candidate. The oval shape with the lower matching degree is left out of the matching candidates. Although the case of the elliptical shape has been described so far, it is possible to evaluate whether or not they are overlapped even in the case of the circular shape. In the case of a circular shape, the size in the X direction and the size in the Y direction are both the number of pixels corresponding to the diameter of the circular shape. In the present embodiment, the threshold for evaluating whether or not they overlap is 10% of the size in the X direction and the Y direction, but the present invention is not limited to this and can be changed as appropriate. it can.

Next, the determination unit 23 calculates a matching score for each of the matching candidates. Specifically, when the matching candidate has a circular shape, as described above, the amount of positional deviation from the pixel that forms the circular shape that is the matching candidate to the pixel that forms the detected edge is acquired. When the acquired positional deviation amount is equal to or smaller than a predetermined value (when the distance corresponding to the acquired positional deviation amount is shorter than the predetermined value), the determination unit 23 stores the pixels constituting the circular shape as matching pixels. To do. Similarly, it is evaluated whether or not all pixels constituting the circular shape that is a matching candidate are matching pixels. The matching score of the matching candidate is calculated by the percentage of the total number of matching pixels and the total number of pixels constituting the circular shape for the circular shape that is the matching candidate. That is, when the total number of matching pixels is the same as the total number of pixels constituting the circular shape (for example, when the detected edge completely matches the geometric shape to be pattern-matched), the matching score is 100% ( Maximum value). Here, the actual edge of the harvested object K is usually not a perfect circle or ellipse, but a shape approximated to a circle or ellipse. For this reason, if the threshold value of the matching score is increased too much, it may be determined that the target is not the harvesting target K to be harvested. In the present embodiment, when the matching score of the matching candidate is 50% or more, the determination unit 23 stores the matching candidate as a matched circular shape M (matched geometric shape M). By setting the threshold value of the matching score as 50%, the harvest target K to be harvested whose actual edge has a shape approximate to a circular shape or an elliptical shape is the edge of the harvest target K to be harvested. Can be expected to be correctly determined.
In the present embodiment, since the measurement unit area W ₁ is not less than the maximum dimension of the crop object K to be harvested dimensions overlapping with the measurement unit area W _2, harvested object K to be harvest, either measurement It will fit in the unit area. However, since the edges of the plurality of harvesting objects K detected by the edge detection unit 22 overlap, some of the actual edges of the harvesting object K may not be detected. Therefore, by setting the threshold value of the matching score to 50%, even if the edges of the plurality of harvesting objects K detected by the edge detection means 22 are overlapped, it is the edge of the harvesting object K to be harvested. It can be expected to be judged correctly. Even if the matching candidate is an elliptical shape, the determination unit 23 can similarly evaluate whether or not the matching elliptical shape M is present.

  A case will be described in which the overlapping dimension of the measurement unit regions is smaller than the maximum value of the dimension of the harvested object K to be harvested. In this case, the smaller the overlapping dimension of the measurement unit regions is, the higher the risk that a part of the harvested object K to be harvested will be removed from the measurement unit region in any measurement unit region. Then, the ratio of the length of the edge of the harvested object K to be harvested in the measurement unit region to the actual length of the edge of the harvested object K to be harvested becomes small. For this reason, if the threshold value of the matching score is increased too much, it may not be determined that it is the harvesting target K to be harvested. Therefore, the threshold of the matching score is set so that even when a part of the harvesting target K to be harvested deviates from the measurement unit region, it can be determined that it is the edge of the harvesting target K to be harvested. It is necessary to lower the value. However, if the threshold value of the matching score is lowered too much, erroneous detection due to noise or the like is likely to occur. However, the smaller the overlapping dimensions of the measurement unit regions are, the smaller the number of times that a series of steps of measuring the harvest target K, determining the harvest target K to be harvested, and cutting the harvest target K to be harvested is repeated. Therefore, the time required for the entire process is reduced. On the other hand, the larger the overlapping dimension of the measurement unit regions, the wider the range in which the harvested object K to be harvested is stored in any measurement unit region. That is, in any measurement unit region, the ratio between the length of the edge of the harvesting target K to be harvested in the measurement unit region and the actual length of the edge of the harvesting target K to be harvested is large. For this reason, even if the threshold value of the matching score is increased to the extent that erroneous detection due to noise or the like is unlikely to occur, it can be expected that the edge of the harvesting target K to be harvested is determined. However, the larger the overlapping dimensions of the measurement unit regions, the more times the above series of steps are repeated, and thus the time required for the entire process increases. In this embodiment, the threshold value of the matching score for evaluating whether or not matching is 50%. However, the present invention is not limited to this, and as described above, it is difficult to cause erroneous detection. In view of the time required for the entire process, that is, based on the overlapping dimensions of the measurement unit regions, it can be changed as appropriate. For example, when there is no overlapping portion of the measurement unit region, in the measurement unit region (XY plane), both the X coordinate and Y coordinate of the harvested object K to be harvested are the X coordinate and Y of the measurement unit region. There is a risk of deviating from the coordinate range. In this case, since the harvest object K to be harvested extends over four measurement unit areas, at least 25% of the edge of the harvest object K to be harvested is included in any measurement unit area. become. For this reason, the threshold value of the matching score is set to a value smaller than 25%.

FIG. 6 shows an example of a circular shape matched to the detected edge. 6A shows the matched circular shapes M1 to M3 and the respective centers C1 to C3, and FIG. 6B shows the matched circular shapes M4 to M6 and the respective centers C4 to C6. In the present embodiment, the determination unit 23 classifies the matched circular or elliptical shapes M into geometric shape groups that are aggregates of circular or elliptical shapes whose centers are close to each other. In the present embodiment, when any of the distances between the centers of the matched circular or elliptical shapes is within a predetermined range, the center positions are circular or elliptical shapes that are close to each other. Specifically, in the present embodiment, the predetermined range, the distance of 5 pixels (e.g., such distance as the position displacement amount d ₃ shown in FIG. 4 (b)) are set as follows. FIG. 6A illustrates a case where all of the distances between the centers C1 to C3 are within a predetermined range. Specifically, the distance between the center C1 and the center C2 (the distance between the center of the pixel located at the center C1 and the center of the pixel located at the center C2), the distance between the center C2 and the center C3 Any of the distances between the center C1 and the center C3 is within a predetermined range. For this reason, the circular shapes M1 to M3 shown in FIG. 6A are classified into one geometric shape group. On the other hand, FIG. 6B shows that the distance between the center C4 and the center C5 and the distance between the center C5 and the center C6 are within a predetermined range, but between the center C4 and the center C6. The case where the distance is not within the predetermined range is illustrated. In this case, the circular shapes M4 to M6 are not classified into one geometric shape group because the center positions cannot be said to be close to each other. Specifically, for example, the circular shape M4 and the circular shape M5 are classified into one geometric shape group, and the circular shape M6 is different from the geometric shape group in which the circular shape M4 and the circular shape M5 are classified. It is classified into the geometric shape group. In the present embodiment, the set predetermined range is equal to or less than the distance of five pixels, but the present invention is not limited to this and can be changed as appropriate.

  And the determination means 23 extracts the circular shape or elliptical shape M with the highest degree of matching among the circular shape or elliptical shape M which comprises a geometric shape group for every geometric shape group. In the present embodiment, for each geometric shape group, a circular shape or an elliptic shape M having the highest matching score is extracted from the circular shape or the elliptic shape M constituting the geometric shape group. When the centers of all matched circular or elliptical shapes M are close to each other, all the circular or elliptical shapes M are classified into one geometric shape group, and the matching score One circular shape or elliptical shape M having the highest is extracted. When the geometric shape group is composed of one circular shape or elliptical shape M, this circular shape or elliptical shape M is extracted as a circular shape or elliptical shape having the highest matching score. Become. And the determination means 23 selects the circular shape or elliptical shape which has a predetermined diameter (diameter of harvesting object) from the circular shape or elliptical shape extracted for every geometric shape group. Specifically, the diameter of the circular or elliptical shape to be selected is determined in advance according to the size of the harvesting target K (mushroom umbrella) to be harvested.

  The determination unit 23 determines that the pixel located at the two-dimensional coordinate of the selected circular or elliptical shape is the edge of the harvesting target K to be harvested in the binarized image formed from the distance image. Specifically, the determination unit 23 selects the selected circular shape or elliptical shape (pixels located at the two-dimensional coordinates of the selected circular shape or elliptical shape in the binarized image formed from the distance image). It is determined (imitation) that the edge of the harvesting target K to be finally harvested. For this reason, whether or not to match the desired shape or diameter in a plane perpendicular to the growth direction, compared with the case where it is determined whether or not to harvest based on whether or not it extends beyond the predetermined height in the growth direction. Therefore, it can be expected to accurately detect the harvest object K to be harvested.

  Further, a circular shape or an elliptical shape having a diameter to be harvested is selected from a circular shape or an elliptical shape extracted for each geometric shape group, and the selected circular shape or elliptical shape has a two-dimensional shape in the distance image. By determining that the pixel located at the coordinates is the edge of the harvesting target K to be harvested, compared to pattern matching only a circular or elliptical shape having a diameter of the harvesting target selected after pattern matching, It can be expected that the edge of the harvesting target K is to be harvested with higher accuracy. Specifically, for example, when the size of the harvesting target K to be harvested is larger than a shape corresponding to a circular shape having a diameter of 40 pixels, the harvesting target K corresponding to a circular shape having a diameter of 37 pixels is used. It is desirable to determine that it is not the harvesting object K to be harvested. When only the circular shape with a diameter of 40 pixels is pattern-matched to the edge detected from the harvested object K corresponding to a circular shape with a diameter of 37 pixels, the circular shape with a diameter of 40 pixels is matched, There is a risk of determining that the edge of the harvesting target K is to be harvested. On the other hand, when a plurality of circular shapes of a circular shape having a diameter of 35 pixels and a circular shape having a size of 40 pixels are subjected to pattern matching with respect to an edge detected from the harvested object K corresponding to a circular shape having a diameter of 37 pixels. In addition, both a circular shape with a diameter of 35 pixels and a circular shape with 40 pixels may match. In this case, it is considered that the positions of the centers of both circular shapes are close to each other. Both circular shapes are classified into one geometric shape group. For one geometric shape group, a circular shape having a diameter of 35 pixels is extracted, assuming that a circular shape having a diameter of 35 pixels has a higher matching degree than a circular shape having a diameter of 40 pixels, It can be expected to more accurately detect the shape of the harvested object K to be harvested. For this reason, it can be expected that the extracted geometric shape is not selected as a geometric shape having a diameter to be harvested. That is, it can be expected that it is not the edge of the harvested object K to be harvested. Thus, it is more accurate to pattern match a plurality of circular shapes of a 35 pixel diameter and a 40 pixel circular shape than to pattern match only a circular shape having a diameter of 40 pixels. It can be expected that the harvesting object K to be determined is determined. In this embodiment, a circular shape or an elliptical shape having a diameter to be harvested is selected from a circular shape or an elliptical shape extracted for each geometric shape group, and the harvest to be harvested by the selected circular shape or elliptical shape. Although the edge of the target K is determined, the present invention is not limited to this, and the edge of the target K to be harvested may be determined based on the matched circular or elliptical shape M.

The signal processing means 2 further includes a cutting position determining means 24. The cutting position determination means 24 is determined to be harvested based on the two-dimensional coordinates and density values of the pixels constituting the pixel region having the edge of the harvested object K determined to be harvested by the determination means 23. The cutting position of the harvested object K is determined. In the present embodiment, the cutting position determination unit 24 extracts a pixel located at the center of the edge of the harvested object K determined to be harvested by the determination unit 23 in the binarized image formed from the distance image. . Then, a three-dimensional position further separated from the three-dimensional position measuring means 1 by a certain distance along the growth direction of the harvested object K from the three-dimensional position corresponding to the two-dimensional coordinates and density value of the extracted pixel is set as the cutting position. decide.
In the present embodiment, the cutting position determination unit 24 selects a pixel located at the center of the edge of the harvested object K determined to be harvested by the determination unit determination unit 23 in the binarized image formed from the distance image. Although the cutting position is determined by extraction, the present invention is not limited to this. For example, the pixels constituting the pixel region having the edge of the harvested object K determined to be harvested by the determination unit 23 The average value of the two-dimensional coordinate and the average value of the concentration value are calculated, and the three-dimensional position measurement is performed along the growth direction from the three-dimensional position corresponding to the calculated average value of the two-dimensional coordinate and the average value of the pixel. A three-dimensional position further apart from the means 1 by a certain distance may be determined as the cutting position.

  As described above, the harvest assistance device 100 according to the present embodiment further includes the cutting means 3 for cutting the harvest target K. The cutting means 3 moves to the cutting position determined by the cutting position determination means 24 and cuts the harvested object K determined to be harvested by the determination means 23. The cutting position is a three-dimensional position measuring means along a growth direction of the harvested object K from a predetermined position of the harvested object K (a three-dimensional position corresponding to the two-dimensional coordinates and density value of the pixel extracted by the judging means 23). Since it is a three-dimensional position further apart from 1 by a certain distance, the length of the harvested object K cut by the cutting means 3 can be made uniform. In the present embodiment, the cutting means 3 is an ultrasonic cutter, but the present invention is not limited to this, and may be, for example, a scissor or a laser cutter. The harvesting target K can be harvested by appropriately collecting the harvesting target K cut by the cutting means 3.

  In the present embodiment, the determination unit 23 performs pattern matching on the edge detected in the binarized image formed from the distance image. However, the present invention is not limited to this, and the distance image Pattern matching may be performed on the detected edge.

  As described above, the present embodiment is characterized in that the harvesting target K to be harvested is determined using the distance image. Then, the test which confirms about the effect by using a distance image was done. Specifically, a pixel having a density value corresponding to the amount of reflected light reflected from the mushroom when the edge of the mushroom that is the harvesting target K to be harvested is determined using the binarized image formed from the distance image A test was conducted to check whether or not the mushroom to be harvested can be determined without problems when the edge of the mushroom to be harvested is determined using the binarized image formed from the grayscale image composed of

  FIG. 7 shows the result when the edge of the mushroom is determined using the gray image of the mushroom. FIG. 7A shows a grayscale image of a mushroom captured by a CCD camera. FIG. 7B shows a binarized image formed from the gray image of the mushroom shown in FIG. FIG. 7C shows a pattern in which a plurality of circular or elliptical shapes having different diameters are pattern-matched with the edges detected by the edge detecting means 22 in the binarized image shown in FIG. 7B. It is a result figure which shows elliptical shape M7. As shown in FIG. 7C, when the edge of the mushroom is determined using the grayscale image, it is determined that the mushrooms indicated by A, B, and C in FIG. The result of not being obtained. This is presumably because mushroom umbrellas overlap each other, or because the density of mushrooms and medium is similar, the difference in density between mushrooms or between mushrooms and medium is reduced.

  FIG. 8 shows a result when a mushroom edge is determined using a mushroom distance image. FIG. 8A shows a mushroom distance image measured by the three-dimensional position measuring means 1 of the present embodiment. FIG. 8B shows a binarized image formed from the mushroom distance image shown in FIG. FIG. 8C shows a pattern in which a plurality of circular or elliptical shapes having different diameters are pattern-matched with the edges detected by the edge detecting means 22 in the binarized image shown in FIG. 8B. It is a result figure which shows elliptical shape M8-M11. As shown in FIG. 8C, it was found that when a mushroom edge is determined using a distance image, a mushroom to be harvested can be determined without problems. This is because the distance image is not affected by the amount of reflected light, so the distance from the three-dimensional position measuring means 1 to the mushroom umbrellas that overlap each other and the distance between the mushroom umbrellas and the medium are as follows: This is because the edges of the mushroom can be accurately determined because of the difference.

  The present invention is not limited to the configuration of the above embodiment, and various modifications can be made without departing from the spirit of the invention. For example, in the present embodiment, for each measurement unit region, a series of steps of measuring the harvest target K, determining the harvest target K to be harvested, and cutting the harvest target K to be harvested are performed. The present invention is not limited to this, and the determination of the harvesting target K to be harvested and the cutting of the harvesting target K to be harvested include, for example, measurement of the harvesting target K growing in each measurement unit region. It may be done after completion. The order in which the three-dimensional position measuring unit 1 moves so that the three-dimensional position measuring unit 1 can measure the two-dimensional coordinates of the harvesting target K and the distance to the harvesting target K for each measurement unit region is as described above. The order is not limited. Further, in this embodiment, the measurement unit region is a range smaller than the medium B, but the present invention is not limited to this, and the measurement unit region may be a range equivalent to the medium B. . Moreover, in this embodiment, although circular shape or elliptical shape is used as geometric shape, it is not restricted to this, Various shapes can be used according to various harvesting objects K.

DESCRIPTION OF SYMBOLS 1 ... Three-dimensional position measurement means 2 ... Signal processing means 3 ... Cutting means 21 ... Distance image formation means 22 ... Edge detection means 23 ... Determination means 24 ... Cutting position determination Means 100 ... Auxiliary harvesting device B ... Medium K ... Harvested object M ... Matched geometric shape

Claims (4)

Three-dimensional position measuring means for measuring the two-dimensional coordinates of the harvested object in a plane orthogonal to the growing direction of the harvested object and the distance to the harvested object along the growing direction in the two-dimensional coordinates;
A harvesting assisting device comprising signal processing means for determining a harvesting object to be harvested,
The signal processing means includes
Distance image forming means for forming a distance image composed of pixels having two-dimensional coordinates measured by the three-dimensional position measuring means and density values corresponding to the distance measured by the three-dimensional position measuring means;
Edge detecting means for detecting an edge of a pixel region composed of the pixels having a density value corresponding to a predetermined distance or less in the distance image;
The detected edge is subjected to pattern matching with a predetermined geometric shape according to the harvesting target to be harvested, and is located in the two-dimensional coordinates of the matching geometric shape in the distance image. A harvesting assisting device comprising: a determination unit that determines that a pixel is an edge of a harvested object to be harvested. The determination means includes
For the detected edge, pattern matching a plurality of circular or elliptical shapes with different diameters as the predetermined geometric shape,
Classify the matched geometric shapes into geometric shapes that are collections of geometric shapes whose centers are close to each other;
For each geometric shape group, a geometric shape having the highest degree of matching is extracted from the geometric shapes constituting the geometric shape group,
Selecting a geometric shape having a predetermined diameter from the geometric shapes extracted for each of the geometric shape groups;
2. The harvest assistance according to claim 1, wherein in the distance image, it is determined that a pixel located at a two-dimensional coordinate of the selected geometric shape is an edge of a harvest target to be harvested. apparatus. A cutting means for cutting the harvest object;
The signal processing means includes
The cutting position of the harvesting object determined to be harvested based on the two-dimensional coordinates and density values of the pixels constituting the pixel region having the edge of the harvesting object determined to be harvested by the determination means Cutting position determining means for determining
3. The harvesting assistance according to claim 1, wherein the cutting unit moves to the cutting position determined by the cutting position determination unit and cuts the harvested object determined to be harvested. apparatus.   The cutting position determining means extracts a pixel located at the center of the edge of the harvested object determined to be harvested in the distance image, and a three-dimensional corresponding to the two-dimensional coordinates and density value of the extracted pixel. 4. The harvest assisting apparatus according to claim 3, wherein a three-dimensional position further spaced apart from the three-dimensional position measuring means by a certain distance along the growing direction from the position is determined as the cutting position.