JPS62250384A - Image tracking apparatus - Google Patents

Abstract

PURPOSE:To discriminate a target in real time with high accuracy and to well track the target by small-sized constitution, by applying correlation processing to high frequency optical signals of picked-up image information and a reference image information etc. CONSTITUTION:Optical signals respectively corresponding to the shape and rotation of an input image from an image pickup part 1 through a converter 2, a zoom device 3, a Fourier transformunit 4, a beam splitter 5, an absolute value device 7 and a mellin transformunit 8, etc. are inputted to an output filter 10 comprising a freely rewritable pervious film through a converter 9. Said filter 10 is controlled corresponding to the reference image information from an optical memory 14 processed by a coordinates converter 15 controlled by an operator 13 to perform the correlation processing of both input and reference images. The correlation result is supplied through an inverse Fourier transformer 11 and a photoelectric converter 12 and an operation means 13 judges whether a target is present. At the time of the tracking of the target, feedback control is performed by the means 13 so that optical correlation information becomes max. and the target can be discriminated in real time with high accuracy by utilizing a high frequency optical signal and good tracking can be performed by a small-sized constitution.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an image tracking device that tracks a target by grasping the target as an image.

(Prior art) In devices that track targets by performing correlation calculation processing on target signals, correlation calculation processing has traditionally been performed using a digital computer, and when a large amount of processing data is processed, a large-sized computer is required. Become. For example, if the tracking device is fixedly installed on the ground and a large, high-speed computer is available, the scale of the device is not so much of an issue, but in the case of complex processing, no matter how fast it is, real-time processing may be difficult. is difficult in most cases.

In addition, when a tracking device is mounted on a flying object and has the function of guiding the flying object to a target, a small, lightweight, and high-speed digital computing unit is required. Currently, processing has to be limited to a limited range due to performance reasons, and it is even more difficult to provide advanced identification functions, including the ability to distinguish between targets and backgrounds, as well as target shapes. This is extremely difficult in terms of the amount of data to be processed and the processing speed.

(Problems to be Solved by the Invention) As described above, with conventional tracking devices, it is difficult to have an identification function that can perform sophisticated real-time processing such as target shape. However, there is a problem in that sufficient tracking accuracy cannot be obtained.

The present invention was made to solve these problems, and it is possible to identify the target shape in real time by taking into account the movement of the target, and it is also suitable for targets that require advanced recognition functions. [Structure of the Invention] (Means for Solving Problems) The image tracking device according to the present invention extracts information on the shape and position information of an input image by correlation calculation using optical signals, and extracts information on the shape and position of the input image. The configuration is such that calculations based on electrical signals are used for the determination and derivation of control signals necessary for control to match the image and the reference image, drive control of the va@ section, and the like.

(Function) With this configuration, it is possible to instantly match the input image and the reference image in response to changes in the image that occur due to changes in the relative distance to the target or the relative orientation of the target, and the input image and the reference image can be matched instantly. It becomes possible to identify shapes and extract their position information in real time.

This makes it possible to achieve sufficient tracking accuracy (q) for targets that require a high degree of identification function.

(Embodiment) An embodiment of the image tracking device according to the present invention will be described below with reference to the drawings. The figure is a schematic configuration diagram illustrating an embodiment of an image tracking device according to the present invention. The R image section (1) captures image information by imaging the outside, and outputs an optical signal corresponding to the image information. Since the optical signal from this pupil image portion (1) is incoherent, it is converted into a cohesin 2~ signal by a converter (2). This transformation is necessary for the Fourier transformation to be performed later, and is performed, for example, by modulating a coherent laser beam with an optical signal from the Il image section (1). The output light of the converter (2) is supplied to a zoom device (3). Zoom device (3)
has the function of enlarging or reducing the input image, and is controlled by the arithmetic means (13).

The output light of this zoom device (3) is converted into a frequency domain optical signal by an optical Fourier transformer (4) and split by a beam splitter (5). One output light of the beam splitter (5) is supplied to the switching device (9) via the absolute value device (γ) and the Melan converter (B), and the other output light is switched via the beam splitter (6). Vessel (9
). The absolute value device (7) discards the position information of the input light and extracts only the amplitude information.

The Melan transformer (8) is a combination of a coordinate transformer and a Fourier transformer, of which the coordinate transformer is composed of a film with, for example, transparency i, which has a pattern change according to a predetermined function, and corresponds to the pattern change. Coordinate transformation is performed by applying intensity changes to input light. Further, the Fourier transformer optically performs Fourier transform on this coordinate-transformed output light. This absolute value device (7) and Melan converter (8)
The part consisting of extracts information on the size and rotation of the input image, and a change in the amplitude of the input light of the absolute value device (7) appears as a change in the phase of the output light of the Melan converter (8).

The switch (9) selects and outputs the output light of either the beam splitter (6) or the Melan converter (8) under the control of the calculation means (13). The output light selected by the switch (9) is supplied to the filter (10) as input image information, where it is multiplied by reference image information. This filter (10) is composed of, for example, a transparent film having a rewritable pattern, and by transmitting the light from the switch (9) through this film, the pattern is used as reference image information and is used as input image information. Perform multiplication. The multiplicative output light of the filter (10) is transferred to an inverse Fourier transformer (11)
The photoelectric converter (12) is optically inverse Fourier transformed by
After being converted into an electrical signal, it is introduced into the calculation means (13). The calculation means (13) controls the entire system and has a high-level judgment function according to a predetermined algorithm.

For example, the target size (
Reference image data according to information such as rotation angle (rotation), direction (orientation, aspect), etc. is stored, and the data is read out under the control of the calculation means (13) and sent to the coordinate converter (13). After coordinate transformation (in this case, coarse rotation) in step 15), the contents of the filter (10) are rewritten and supplied as a signal for controlling ilJ. That is, the coordinate converter (15) enlarges, reduces, or roughly converts the reference image data read from the optical memory (14) based on the coordinate conversion data regarding the size and rotation of the image given from the calculation means (13). It gives the rotation. The coordinate conversion function of the coordinate converter (15) and the zoom device (3) allows the relative distance to the target and the eye to be determined.

The image tracking device constructed in this way operates as follows. That is, in order to grasp and track the target image, the operation starts by first determining whether or not the target exists within the image. For this purpose, the switch (9) is controlled to select the output light of the Melan converter (8) by a command from the calculation means (13). As one method for determining the presence or absence of a target, a Melan converter (8), a filter (10),
One example is a method in which the calculation means (13) determines that a target exists when the correlation output level obtained by the correlation calculation of the inverse Fourier transformer (11) exceeds a certain threshold level.

When it is determined that the target exists, the zoom device (3) matches the shape of the coordinate-changed image on the screen under instructions from the calculation means (13). This matching judgment is made by the calculation means (
13), and the optical memory (
14), the zoom device (3), and the coordinate converter (15) are controlled, but since this is done at the speed of light, matching is actually achieved instantaneously. Zoom device (3) and coordinate converter (
The coordinate transformation according to 15) is performed in order to prevent a decrease in correlation efficiency caused by a scale change in which the size of the image changes without changing its shape, or a rotation change in which the image is rotated with respect to the center of coordinates. That is, by controlling the Melan converter so that it rotates at a constant rate, the above-mentioned decrease can be prevented.

In addition, the direction of the image (orientation, aspect)
Even if , the correlation efficiency decreases. In this case, it is possible to prevent the efficiency from decreasing by using multiple matched filters and combining a plurality of matched filters.

If it is determined that the target exists in this way, the calculation means (
13) controls the switch (9) to select the output light of the beam splitter (6) and performs a correlation calculation for calculating the target position. In this case, optical memory (1
The read data from 4) is also switched and filtered (10
) has already been rewritten into image information in the frequency domain for detecting the target position, so the correlation output of the inverse Fourier transformer (11) is maximum and its peak value is very sharp. Become. As a result, target position information can be extracted with extremely high accuracy, and the calculation means (
13), for example, determines that the position of the correlation output value equal to or higher than a certain threshold 1 hold level is the target position, and uses this position information to cause the drive system in the peripheral direction of @image section (1) to track the target. control.

In reality, the switch (9) is switched alternately at high speed, and correlation calculation processing by Melan transform and correlation calculation processing for extracting target position information are performed alternately. Even if the shape changes, it is possible to track the fill in real time by looping from the optical memory (14) to the filter (10).

By the way, the output light from one side of the beam splitter (6) is converted into an electrical signal by a photoelectric converter (1G) and sent to the calculation means (13).
Since this signal is an input image signal in the frequency domain, by appropriately performing processing such as segmentation and clustering in the calculation means (13), it is possible to identify and detect the background and the target. It is also possible to obtain information such as the degree of spread of the target and its direction, and it is effective to take this information into account when making various decisions.

The image tracking device shown in the figure is capable of identifying the shape of a target by considering the relative movement between the target and the tracking device, and is capable of identifying and selecting the optimal target from multiple targets. It is suitable for anti-tank systems that require image processing capability, for example.

Furthermore, by fixing the reference image information of the filter (10), it is possible to track only a target coming from a predetermined direction, for example, only when there is an input image that matches the reference image information. .

Roughly speaking, in this image tracking device, the basic image processing that requires high speed is performed by the optical system, and the processing of algorithms that require complex judgments is performed by the electrical calculation processing of the digital processor, so that the entire device is Not only can it be made smaller, but since the main story processing section is comprised of an optical processing section, it is resistant to interference and damage caused by radiation and electromagnetic waves.

[Effects of the Invention] As explained above, according to the image tracking device according to the present invention, the target shape can be identified in real time by taking into account the relative movement of the target, and the target can be tracked with high precision. It is possible to reduce the size of the device, and it is also excellent in terms of resistance to radiation and electromagnetic waves, and has great practical effects.

[Brief explanation of drawings]

The figure is a schematic configuration diagram illustrating an embodiment of an image tracking device according to the present invention. (1)...imaging section, (4)...Fourier transformer. (7)... Absolute value device, (8)...
・Melan converter. (9)...Switcher, (10)...Filter, (11)...Inverse Fourier transformer,
(12)...Photoelectric converter, (13)...
...Arithmetic means, (14)...Optical memory, (
15) Coordinate converter. Agent Patent Attorney Nori Chika Kenshu Kanshu Kikuo Takehana

Claims (1)

[Claims] An imaging means that takes in external image information, and an optical correlation that is connected to the imaging means and extracts a first correlation output corresponding to information regarding the shape of the input image and a second correlation output corresponding to the position information of the input image. means, which is connected to the optical correlation means, matches the input image and the reference image in terms of shape based on the first correlation output, and determines the imaging direction of the imaging means based on the second correlation output; An image tracking device comprising a control means for controlling an image.