Method for the characterization of objects

Abstract

A method for characterizing an object using distance measurement includes: determining elevation profiles using distance measurement, and evaluating the determined elevation profiles for a characterization of the object, the characterization includes a position and/or at least one object-specific parameter of the object.

Claims

1. A method for characterizing an object using distance measurement, the method comprising: a) determining elevation profiles based on a distance measurement, b) evaluating the determined elevation profiles for a characterization of the object, and c) determining the position of the object, wherein the characterization comprises at least one of a position and at least one object-specific parameter of the object, wherein surface elements are determined based on the determined elevation profiles, wherein normal vectors are determined for each of the surface elements, wherein the normal vectors originate from the respective surface element, wherein a point of intersection, a point having a highest intersection density, or a mean value of all points of intersection of the normal vectors is determined as the position of the object, wherein a region B with an area A is determined, within which the object is located, and wherein a maximum length of the normal vectors is limited to a region B with an area C.sub.1.sup.2.Math.A which is arranged symmetrically with respect to the origin of the normal vector, where C.sub.1 is greater than 1.

2. The method of claim 1, wherein the region B with the area A is determined, within which the object is located, and wherein a minimum length of the normal vectors is limited to a region B with a surface C.sub.2.sup.2.Math.A which is arranged symmetrically with respect to the origin of the normal vector, where C.sub.2 is smaller than 1.

3. The method of claim 1, wherein a length of the normal vectors is modified based on at least one of a distance from a center of the object and an angle between the perpendicular and the respective normal vector.

4. The method of claim 1, wherein the object comprises a harvest product.

5. The method of claim 1, wherein the distance measurement is effected using ultrasound or laser triangulation.

6. The method of claim 1, wherein the evaluation of the determined elevation profiles for the characterization of the object is performed in real time.

7. The method of claim 1, wherein the distance measurement is performed using a sensor, and the object is moved relative to the sensor by moving the object and/or by moving the sensor.

8. The method of claim 1, wherein a continuous evaluation is performed after each newly detected elevation profile.

9. A method for characterizing an object using distance measurement, the method comprising: a) determining elevation profiles based on a distance measurement, b) evaluating the determined elevation profiles for a characterization of the object, and c) determining the object-specific parameter of the object, wherein the characterization comprises at least one of a position and at least one object-specific parameter of the object, wherein a region B with an area A is determined, within which the object is located, wherein a histogram is generated for all data points of the elevation profiles located within the region B, wherein the histogram is divided into a lower section, an intermediate section and an upper section, wherein a weighted mean value is determined for the intermediate section of the histogram, and wherein a median of the histogram divided by the weighted mean value is determined as a relative density value.

10. A method for characterizing an object using distance measurement, the method comprising: a) determining elevation profiles based on a distance measurement, b) evaluating the determined elevation profiles for a characterization of the object, and c) determining an object-specific parameter, wherein the characterization comprises at least one of a position and at least one object-specific parameter of the object, wherein a region B with an area A is determined, within which a harvest product is located, wherein a number of the data points within the region B is determined as Ntotai, wherein the area A of the region B is determined, wherein a radius R is defined, wherein a number of the data points within the radius R is determined as N.sub.core, and wherein a ratio of (A.Math.N.sub.core)/(R.sup.2.Math..Math.N.sub.total) is determined as an opening parameter.

Description

DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows a schematic set-up for distance measurement,

(2) FIG. 2 shows a line-wise elevation profile determined by distance measurement,

(3) FIG. 3 shows a detail of three elevation profiles with the surface elements determined,

(4) FIG. 4 shows a heat map for indicating of the intersection density of normal vectors of the surface elements,

(5) FIG. 5A an exemplary surface element with a normal vector,

(6) FIG. 5B shows a functional relationship between the modification of the length of the normal vectors and the angle between the perpendicular and the normal vector;

(7) FIG. 6 shows a histogram of the determined data points of the distance measurement,

(8) FIG. 7A is a side view of a harvest product in the form of lettuce in an open state,

(9) FIG. 7B shows the harvest product of FIG. 7A in top plan view,

(10) FIG. 8A is a side view a harvest product in the form of lettuce in a closed state, and

(11) FIG. 8B shows the harvest product in top plan view.

DETAILED DESCRIPTION OF THE INVENTION

(12) In a method for characterizing an object by means of distance measurement, line-wise elevation profiles are first determined by means of distance measurement and then evaluated in a suitable manner for the characterization of the object, wherein the characterization of the object includes the position and/or at least one object-specific parameter of the object. In this context, a laser triangulation system as illustrated in FIG. 1 can be used for distance measurement. The system comprises a sensor 10 in which a line laser 12 is arranged which illuminates an object 16 via a mirror 14. Via a further mirror 20, the line generated by the laser 12 on the object 16 is imaged by a camera 18 which is also arranged in the sensor 10. The line on the object 16 is distorted by the topology of the object 16. Based on this distortion, an evaluation means which is connected to the camera, can determine an elevation profile along the line generated by the laser 12. Thereby, according to the coordinate system in FIG. 1, the Z-component of the object 16 is determined. Such an elevation profile 22 is illustrated e.g. in FIG. 2 and is composed of a great number of data points 24 per line, in particular 500 to 5000 data points per elevation profile. Thereafter, for the complete detection of the object 16, either the object and/or the sensor 10 is displaced along the Y-axis, and another line-wise elevation profile is determined as described before. Thereby, a large number of elevation profiles is determined. These typically have a mutual distance of 0.5 mm to 2 cm, where the distance is in particular constant and is predetermined e.g. by a continuous movement of the sensor 10 or the object 16 relative to each other until a time has lapsed, or after a relative movement of the sensor 10 by a predetermined distance with respect to the object 16.

(13) FIG. 3 illustrates a detail showing a part of three elevation profiles 22. Here, the points 24 indicate the data points of the respective elevation profile. For the determination of the position of an object, surface elements 26 are determined from the elevation profiles 22. In FIG. 3, these are triangles. However, other surface elements are also conceivable. The corners of the surface elements 26 are located on the data points 24 of the elevation profiles 22 and are connected by the sides of the surface elements. This process is also known as tessellation.

(14) For each of the surface elements 26 a normal vector 28 is then calculated which stands perpendicularly on the surface element 26. For reasons of clarity, FIG. 3 only shows one of these normal vectors 28. Thereafter, all points of intersection of these normal vectors 28 are determined. FIG. 4 illustrates a representation in the form of a heat map as an example, wherein the position of the points of intersection has been projected onto the x/y-plane and the density of the points of intersection is shown in the heat map of FIG. 4. Here, the projection is made into the plane parallel to the plane in which the objects are arranged. The site with the highest intersection density 30 is assumed to be the position of the object 16. FIG. 4 illustrates another object 32 which comprises another site of the highest intersection density 34. Thereby, the position of the two objects 16 and 32 is known. This is true in particular irrespective of the shape of the object as long as the same has a substantially concave or convex surface, wherein, in case of a concave surface, the normal vectors are defined such that they are directed towards the sensor 10. Whereas, in case of a convex or substantially convex shape of the object 16, 32, the normal vectors are defined such that they are directed away from the sensor 10, i.e. in the negative Z-direction.

(15) Thus, a robust method is obtained for determining the position of an object, in particular for similar, but not exactly identical objects, e.g. plants, animals, cars and the like.

(16) For a reduction of the computational effort for the determination of the points of intersection of the normal vectors, a radius R is determined within which the object 16, 32 is essentially located. For the purpose of illustration, the radius R is drawn for the object 16 in FIG. 4, represent by the circle 36. The maximum length of the respective normal vectors 28 can now be limited to C.sub.1.Math.R, where C.sub.1 is chosen to be larger than 1. As an alternative or in addition, the minimum length of the normal vectors 28 can be limited to C.sub.2.Math.R, where C.sub.2 is chosen to be smaller than 1. In this manner, both remote points of intersection and points of intersection immediately adjacent the surface element are not considered in the determination of the position of the object. This is also illustrated in FIG. 5A which illustrates a single surface element 26 as an example. A normal vector 28 originates from this surface element 26. A first region 31 is arranged around the surface element 26 or the origin of the normal vector 28, which region defines the minimum length of the relevant normal vector 28. A second region 33 is defined around the surface element 26 or the origin of the normal vector 28, which region indicates the maximum length of the relevant normal vector 28. Thus, only the region 35 of the normal vector 28 that is located between the regions 31 and 33 is taken into account. This is done for each of the normal vectors 28 determined. In this manner, only that section of the respective normal vectors 28 is considered in which an intersection with another normal vector 28 can be expected. This allows for a significant reduction of computational effort, so that the method can be executed faster, in particular in real time.

(17) Further, a modification of the length of the normal vector can be considered based on the angle a between the normal vector and the vertical z-direction. Here, the modification of the length of the normal vectors to be considered can be effected based on an empirical dependence on or, as illustrated in FIG. 5B, based on a functional relationship, wherein the modification is effected via a parameter C.sub.a plotted over the angle .

(18) Thereby, the computational effort for the determination of the points of intersection of the normal vectors can be maintained low and in particular it is possibly to guarantee a determination of the position of an object in real time.

(19) For the determination of a relative density center of the object as an object-specific parameter, the data points within a region are plotted as a histogram, as illustrated in FIG. 6. Thereafter, the histogram is divided into a lower section 38, an upper section 40 and an intermediate section 42 situated therebetween. Here, for example, starting from the smallest elevation values or z-values, 70% of the data points are located in the lower section 38 and 10% of the data points are located in the upper section 40, starting from the greatest elevation values or z-values. Thus, in the example illustrated, the intermediate section 42 inbetween comprises 20% of the data points. Other distributions are also possible, where, for example, the lower section 38 can comprise in particular 50% to 90% of the data points and the upper section 40 can comprise 2% to 30%. A weighted mean value is calculated for the values of the intermediate section 42. In addition, a median of the histogram is determined and the median is divided by the weighted mean value of the intermediate section 42. The result is assumed as the relative center of density which corresponds to the relative center of gravity of the object. Here, depending on the kind of object characterized by the method, the region of the lower section 38 and of the upper section 40 can be adjusted such that the relative center of density thus determined coincides with the actual center of density.

(20) The method for determining an opening parameter will be described hereunder using an object in the form of lettuce as illustrated in FIGS. 7A to 8B.

(21) FIG. 7A is a side view of a lettuce in an open state. Most leaves 44 of the lettuce grow in a substantially vertical direction or z-direction. A top plan view of the lettuce 43 is illustrated in FIG. 7B. FIG. 8A schematically illustrates a lettuce 46 in a closed state in which the leaves have grown together in the center of the lettuce 46 and thereby form an in particular spherical surface 48. A circle 50 is defined with a radius R within which the harvest product and in particular the lettuce 43, 46 is located. Thereafter, the number of data points present within the circle 50 is determined as N.sub.total. Next, a second circle 52 is defined with a radius R around the same center as the circle 50. In this regard, the radii R and R and the ratio between the radii R and R can be chosen freely and can be adapted to the respective harvest products, depending on the application. In particular, the radii in FIG. 7A and FIG. 7B can be chosen to be different.

(22) Thereafter, the number of data points within the circle 52 is determined as N.sub.core. The ratio of N.sub.core to N.sub.total relative to the surface of the respective circle 50, 52 is calculated as the opening parameter. Thus, the opening parameter is defined as the ratio between the data point densities within the circles 50 and 52. In other words: if the ratio between the radii R and R is indicated by C, where 0C1, the opening parameter is obtained as N.sub.core/(C.sup.2.Math.N.sub.total).

(23) Since the leaves of the lettuce 43 are essentially parallel to the sensor 10, no or only a small number of data points can be determined for the same. On the other hand, a large number of data points can reliably be determined in the spherical core region of the lettuce 46, just because the spherical surface 48 of the lettuce 46 is oriented perpendicularly to the sensor and can thus be easily detected by the same. In this manner, an object-specific parameter can easily be obtained which indicates the state of the harvest product and thus allows for a reliable performance of a subsequent treatment or harvest. Here, the individual shape of the specific harvest product or of a single lettuce or cabbage can be taken into account.