JPS63155366A - Forming system for cubic form - Google Patents

Abstract

PURPOSE:To form an overall cubic form from plural sheets of distance images by showing each distance image in the form of an octo-tree and obtaining a product set of octo-trees. CONSTITUTION:A distance image input part 1 is provided together with an octo-tree showing part 2 and a cubic formation part 3. The part 1 serves as a part where plural distance images of an object are inputted when this object is viewed from the prescribed direction. Then each distance image is shown in the form of an oct-tree by the part 2. The part 3 obtains the product of all distance images shown in the octo-tree forms. Thus a cubic form is obtained. In this case, a spot or slit projection method is applied to obtain those distance images and these images are shown in octo-tree forms independently of each other. Then a cubic form is obtained with no interpolation, etc., and therefore the time efficiency is improved.

Description

[Detailed Description of the Invention] [Summary] In three-dimensional shape measurement, the present invention expresses each distance image as an octree in order to construct an entire three-dimensional shape from a plurality of distance images, and then calculates the intersection of the octrees. By obtaining this, we construct the shape of an object.

[Industrial application field]

The present invention relates to a three-dimensional shape construction method in three-dimensional measurement,
In particular, the present invention relates to a three-dimensional shape construction method in which an octree shape is obtained from a plurality of distance images and a three-dimensional shape of an object is constructed using the octree shape.

[Problems to be solved by the prior art and the invention] Conventionally, industrial robots have been used to measure the distance to objects using vision;
Three-dimensional measurement technology is used to measure relative positional relationships, obtain three-dimensional shapes, etc. Representative methods of three-dimensional measurement include spot projection method and slit projection method. In either method, beam light or slit light is obtained using a projector, and the reflected light is measured by rotating the television camera.The position of the reflected light from the target object projected onto the television camera is then measured by a computer. Through analysis, the distance to the object, positional relationship, three-dimensional shape, etc. are obtained.

In this way, the shape measurement of three-dimensional objects is basically done without contact, and in the spot projection method, the distance of each point obtained is measured by triangulation and three-dimensional linear interpolation is performed to measure the entire shape. The three-dimensional shape is obtained. The slit projection method is an improved version of the spot projection method, in which the spot light is replaced with a slit light, which improves the time required for distance measurement.

However, in any of these methods, in order to obtain the three-dimensional shape of a three-dimensional object, it is necessary to measure the distance to the object, and there is a problem that it takes time, so an efficient method for obtaining the three-dimensional shape is desired.

[Means and effects for solving the problems] The purpose of the present invention is to provide a three-dimensional shape construction method for three-dimensional measurement that solves the above-mentioned problems, and FIG. 1 shows a block diagram of the principle of the present invention. Basically, the distance image input section 1
The distance image input unit 1 is a part that manually generates multiple distance images of an object when viewed from a predetermined direction, and the octree expression unit 2 inputs each distance image. The distance image is expressed in an octree shape, and the three-dimensional configuration unit 3 calculates the product of all the distance images represented in the octree, thereby obtaining the three-dimensional shape. In this case, distance images are obtained by using the spot projection method or slit projection method described above, but after expressing the obtained multiple distance images independently as an octree, a three-dimensional image is constructed without interpolation. Therefore, time efficiency can be achieved.

〔Example〕

FIG. 2 is a block diagram of an apparatus for implementing the three-dimensional shape construction method according to the present invention. In FIG. 2, 100 is a distance image input section, 200 is an octree representation section, 301 is a maximum node level detection section, 302 is a node sharing detection section, 30
3 is a node verification unit, and 304 is a memory. 301-3
04 constitutes a three-dimensional component.

The distance image input unit 100 identifies the object as A, as shown in FIG.
This is a commonly used distance measuring device that inputs a plurality of distance images when viewed from directions B and C. That is, any method for obtaining distance information of a three-dimensional object may be used, for example, the method of projecting light using spot light or slit light as described above may be used.

The octree representation unit 200 is a part that represents an input distance image as an octree. The octree representation, as shown in Figures 4 and 6, is one in which eight trees are sequentially obtained, and it shows the state occupied by the target object in a space divided into eight parts. ing. That is, in Figure 3, the view seen from the A direction (A view) is shown in A, the B view is shown in B, and the C view is shown in C. Considering a cube as shown by the dotted line, In this case, as shown in FIG. 4, for each of the eight cubes indicated by a number, white dots, black dots, and half-black dots indicate whether or not they are occupied by an object, which is an octree representation.

In this case, a white point is occupied by space and no object exists, and a black point is a state in which all space is occupied by an object.
A half-sunspot indicates that a portion of the image is occupied by an object. If we look at Figure 4(a), in the first state, the object occupies a part of the space, so the first level is a half-sunspot, and the specific position occupied by the object with respect to this half-sunspot is Since it is a cube numbered 3.6.7, 3.6.7 on the second level is a black dot. In this case,
Regarding number 7, since it is a shadow in the A view, it is marked as a black dot as an object exists, regardless of whether it is actually occupied or not.

Next, the octree shown in FIG. 4(b) will be explained. In this case, as before, the first level is a half-sunspot, but in the second level, numbers 3 and 6 are occupied by objects, and number 7 is occupied by objects in the B view. Since the object occupies only a portion of the area, it becomes a half-sunspot. Next, for this number 70 half-black point, 8 divisions are provided as described above, and which number division is occupied by an object is marked at the next level. Thereafter, in the same manner, stages are created one after another for the half-sunspot locations, and the tree grows downward. That is, the tree grows downward in the portion numbered 7 in FIG. 4(b), and the tree grows in the portion numbered 6 in FIG. 4(c). Even if the shape is complex, octree representations can be developed one after another in this way.

Figure 4 (a), (b), (C
), find the intersection set by multiplying them. That is, find Al''IBnc.

In this case, each mark has a priority order, with white dots, half-black dots, and black dots having higher priorities in this order. When actually obtaining the intersection set, for example, Al"1Bnc is a white point x a half black point
When it is a half-sunspot, the white point is used; when it is a half-ring dot x sunspot x sunspot, the half-sunspot is used. Furthermore, if all the points are half-sunspots, half-ring points are naturally adopted. In this way, Figure 4 (,
The product set AnBnC, which is obtained by multiplying each level of l) and (bL (c), is obtained as shown in Figure 5. From the octree representation obtained in this way, it is traced in the exact opposite way to the above. By doing this, the specific three-dimensional shape shown in FIG. 6 can be obtained.

The maximum node level detection unit 301 shown in FIG. 2 determines the maximum node level of all distance images expressed as octrees. The uppermost half black dot in FIGS. 4 and 5 is the maximum node level referred to here.

As is clear, the maximum node level is shared by white and black. This indicates that the object is located in a portion of the 8 divided spaces as described above. The node sharing detection unit 302 detects node sharing locations for each level in order from the maximum node level. For example, this corresponds to the half black dot number 7 in FIG. 4(b). The node detecting unit 303 writes the complete nodes among the node sharing parts into the memory 304, and returns the incomplete nodes to the node sharing detecting unit 302 again. This series of processing ends when the node level reaches the lowest level.

〔Effect of the invention〕

As described above, according to the present invention, the three-dimensional shape of an object can be measured from a plurality of distance images regardless of the method of distance measurement, so there is an effect of efficient measurement. In addition, since the three-dimensional structure is memorized using octree representation,
There are also effects such as ease of data transmission.

[Brief explanation of the drawing]

Figure 1 is a block diagram of the principle of the present invention. Figure 2 is a block diagram of a device that implements the three-dimensional shape construction method according to the present invention. Figures 3 (a), (b), and (c) explain the three-dimensional shape. Figure 4 (a), (b), and (c) are diagrams explaining octree representation, Figure 5 is a diagram explaining octree obtained by intersection set, and Figure 6 is diagram 5. FIG. 2 is a diagram illustrating a three-dimensional shape obtained by the octree. (Explanation of symbols) 1.100...Distance image input unit, 2.200...Octree expression unit, 3...3D configuration unit, 301...Maximum node level detection unit, 302...Node Shared detection unit, 303... Node detection unit, 304... Memory.

Claims (1)

[Claims] 1. In a three-dimensional shape construction method in three-dimensional measurement, there is a distance image input unit that inputs distance images when the target object is viewed from a plurality of predetermined directions, and an octograph for each of the plurality of directions from the obtained distance images. an octree representation unit that obtains a tree, a three-dimensional construction unit that obtains an intersection set at each node level of each of the octrees, and further obtains an octree from the intersection set, based on the octree obtained from the intersection set. A three-dimensional shape construction method characterized in that a three-dimensional shape is obtained by using three-dimensional shapes.