JPH0468478A - Picture processor - Google Patents

Abstract

PURPOSE:To automatically identify a target object moving in a front direction or the like by setting a wide window in the range of wide view including the target object, identifying the target object according to the output of symmetry evaluation in a target window scanning the inside of this wide window, and executing initial setting. CONSTITUTION:Corresponding to the output of a self-vehicle advancing direction angle detector 13, a window setter 12A traces a target vehicle 5 in an advancing direction and automatically sets a wide window 14 of wide view including the trace target vehicle 5, and based on the operation of the window setter 12A, the inside of the wide window 14 is scanned by a target window 18 having (m) rows and (n) columns. The inside of the wide window 14 is continuously scanned by the target window 18 until a picture in the target window 18 comes at the most symmetrical position with the central axis of the target window 18 as a reference, and the object 5 is identified and initially set. Thus, the target object 5 can be automatically identified.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing apparatus, and more particularly to an image processing apparatus capable of automatically identifying a target object.

[Prior Art] Examples of conventional image processing apparatuses capable of defining a distance include those shown in Japanese Patent Publication No. 3805/1983, Japanese Patent Publication No. 46363/1983, and the like. These conventional examples are
This will be explained using figures. FIG. 8 is a configuration diagram showing an example of a conventional image processing device, and in FIG.
) is a pair of left and right optical systems, which are placed apart by a baseline length B. (3) and (4) are image sensors, which are installed at the focal length f of the optical systems (1) and (2). (
5) is an object, which is imaged onto image sensors (3) and (4) by optical systems (1) and (2). (6),
(7) is an A/D converter and image sensor (3), (
4) Convert the above image signal to a digital signal. (8),
(9) is a memory in which the outputs of the A/D converters (6) and (7) are stored. (10) is a microprocessor which processes image signals stored in memories (8) and (9) to measure the distance to the object (5). (11
) is a display, and (12) is a window setting device for manually setting a window for the target object (5).

Next, the operation will be explained. The image signal of the target object (5) that has entered the window is stored in memories (8) and (9). And the microprocessor (10) has a memory (8
) and (9), and compares the image signal on the image sensor (3) with the image signal on the image sensor (4) by shifting the image signal on either image sensor. Do it while doing it.

In this way, the shift amount of the image signal that best matches the two image signals is detected, and if this shift amount is a, and the pitch of the pixels of the image sensors (3) and (4) is p, then the object (5) The distance to L-fx is determined by the principle of triangulation.
It is calculated as B.

X p Also, a method of continuously tracking a specific image signal was developed in
It is disclosed in Japanese Patent No. 0-33352. This method is configured so that the operator can arbitrarily set a window with a similar shape to the image signal to be tracked on the display screen while looking at the display screen. Its distinctive feature is that it is designed to prevent people from entering as much as possible.

[Problems to be Solved by the Invention] The conventional image processing device is configured as described above, and combines a distance detection device and an image tracking device to continuously track a specific image signal and calculate the distance to the image. Therefore, there were safety problems if the driver set complicated windows while looking at the display screen while driving the vehicle.

The present invention was made to solve these problems, and an object of the present invention is to provide an image processing device that can automatically identify a target object moving in front of the object.

[Means for Solving the Problems] An image processing apparatus according to the present invention includes means for comparing left and right image signals with one of the left and right image signals as a reference to determine a distance, and a means for calculating a distance by comparing the left and right image signals, and for calculating a distance at a predetermined time interval. Compare the image signal within a sampled time window with the image signal at the next time, detect the part where the image signals match best, set this part as the target window, set the reference image signal, and measure the change in the image signal. In an image processing apparatus having means for determining the moving direction and amount of movement of an object from the amount, means for setting a wide window in a large field of view including the object near the front, and scanning a target window within the wide window. and symmetry evaluation means for evaluating the symmetry within the target window using at least one optical system, and the target object is identified and initialized based on the output of the symmetry evaluation means.

[Operation] In this invention, a wide window is set in a wide field of view including the target object in front, and the target object is identified by the output of the symmetry evaluation within the target window that scans within this wide window. , is initially set.

[Embodiment 5] Hereinafter, an embodiment of the present invention will be described with reference to the drawings, taking as an example the case where the target object is a car. FIG. 1 is a block diagram showing one embodiment of the present invention, and (1) to (11) are completely the same as the conventional device described above.

In this embodiment, a window setting device (12^) and an own army advancing direction angle detector (13) are provided, and the wind setting device (12^) is set according to the output of this own army advancing direction angle detector (13).
The window can be set automatically.

Next, the target identification operation will be explained with reference to FIGS. 2 to 6. The window setter (12^) tracks the target vehicle in the advancing direction as shown in Fig. 2 according to the output of the self-progressing direction angle detector (13), and the wide window (14) provides a wide field of view that includes the tracked target vehicle. is automatically set as shown in Figure 3. This wide window (14) changes according to the output of the own army advancing direction angle detector (13) so as to always capture the field of view in the advancing direction of the own army (Fig. 4).
15) shows the shape of the wide window when the vehicle is traveling straight, and when the vehicle is steered to the right or left, the wide window changes to a shape that corresponds to the field of view in the direction of travel, as shown in (16) and (17), respectively. Then, set the window setting device (12Δ) inside the wide window above.
Under the operation of , a target window (18
) is scanned (Figure 5). At this time, in the target window (18), focusing on the fact that the object (5) is a car and is generally symmetrical,
8) Continue scanning the wide window (14) with the target window (18) until the image in the target window (18) reaches the most symmetrical position with respect to the central axis of the target window (18) (Figures 5 and 6) ).

In this way, the object (5) is identified, and then the target window (]8) captures the image of this object (5) and the image remains symmetrical with respect to the central axis of the target window (18). Target window (1
8) Constantly change the position of. At this time, the target window (18) may appear from the wide window (14).

An example of the above symmetry evaluation method will be explained using FIG. 6. In Fig. 6, Fig. 6(a) is a reference screen, Fig. 6(b) is a screen later shifted from the reference screen at time t, and Fig. 6(e) is a final symmetrical screen. represents.

T(Δs): ΣΣIL(ps+Δs+i, qu+j)−
L(ps+m+Δs−'+q”j) Here, T(ΔS) represents the evaluation degree of symmetry. After scanning ΔS within a predetermined range, T(ΔS) is calculated by the microprocessor (10), and the most Find the position ΔS where the value of T(ΔS) is small, and use the area where the upper left pixel position of the target window (18) is (ps+Δs, qu) and the size is mXn as the reference image, and use it as the target window (18). Reset.Furthermore, in the same way as at point 1-70, transfer the area specified as the target window (18) in the memory (8) to the target window memory.This target window memory is naturally the memory (8). , (9), may be anywhere in the memory that the microprocessor (10) has.

Next, inter-vehicle distance measurement will be explained with reference to FIG. 7. The microprocessor (10) scans the reference image signal surrounded by the target window (18) within the area (19) of the right image captured by the image sensor (4) in order to calculate the distance to the object (5). to find the most matching scan position. As this detection method, for example, the pixels of the left image are indicated by L(i, j), the pixels of the right image are indicated by R(i, j), and the size of each image is x! , Target Wind (18
) is represented by the upper left pixel L (p, q), and at this time, the upper left position of the area (19) of the right image is (1, q), which is expressed by the following formula, where S is the scanning position of the right screen. Perform the calculations that will be performed.

P(s)□Σ ΣIL(9"!, Q"j>-R(s+i
, q+j) Here, scan S from 0 to -m and then P
(s), and the smallest P (s) value is 1 ms.
seek. That is, this scanning position ms indicates the position of the image in the right screen that most matches the reference image. This ms
corresponds to rr shown in FIG. 1, while p corresponds to 11. Therefore, the distance to the object (5) specified using the target window (18) is the optical system base length B,
If the focal length is f and the pixel pitch of the image sensors (3) and (4) is P, then it can be determined as XB L''.

Next is the image tracking method, which uses the target window (1
Since the image of the object (5) is always at the center of the window 8), the direction of the object (5) is determined by the deviation between the center of the target window (18) and the optical axes of the optical systems (1) and (2). Can calculate.

Furthermore, by processing this direction in time series, the moving direction and moving speed of the object (5) can also be detected.

[Effects of the Invention] As described above, according to the present invention, there is provided a means for comparing left and right image signals with reference to one of the left and right image signals to determine a distance, and a means for sampling the image signal at a predetermined time interval. Compare the image signal within a window at a certain time with the image signal at the next time, detect the part where the image signals match best, set this part as the target window, set the reference image signal, and calculate from the amount of change. In an image processing apparatus having means for determining the moving direction and amount of movement of a target object, a means for setting a wide window in a large field of view including the target object near the front, and scanning the target window within the wide window; A symmetry evaluation means for evaluating the symmetry within the target window using at least one optical system is provided, and the target object is identified and initialized based on the output of the symmetry evaluation means, so that the driver's device can be easily set. This has the effect of eliminating the need for operations and ensuring safety.

[Brief explanation of the drawing]

Fig. 1 is a configuration diagram showing an embodiment of the present invention, Figs. 2 to 5 are explanatory diagrams of predicting and identifying the position of an object, Fig. 6 is an explanatory diagram of symmetry axis detection, and Fig. 7 is an explanatory diagram of the detection of the axis of symmetry. An explanatory diagram of distance measurement, and Fig. 8 is a configuration diagram showing a conventional image processing device.6 In the figure, (1) and (2) are optical systems, (3) and (4)
is the image sensor, (8), (9) is the memory, (10)
is a microprocessor, (12^> is a window setting device,
(13) is a self-army advancing direction angle detector. In addition, in the figures, the same reference numerals indicate the same or corresponding parts. Agent Zeng Ido Figure 2] 4 Figure 4 t7 IF) Figure V left Ih 11] Procedural Amendment September 19, 1991

Claims (1)

[Claims] Means for determining a distance by comparing left and right image signals with one of the left and right image signals as a reference; Compare the image signal at the time of , detect the part where the image signal most closely matches, set this part as the target window as the reference image signal, and calculate the moving direction and amount of the object from the amount of change. means for setting a wide window in a large field of view including the object near the front, scanning the target window within the wide window, and at least one optical system for scanning the target window within the target window. 1. An image processing device comprising: symmetry evaluation means for evaluating the symmetry of a target object; and an object is identified and initialized based on the output of the symmetry evaluation means.