SU1737755A1 - Device for finding and determining object coordinates in image - Google Patents

Abstract

The invention relates to a technique of digital image processing and can be used in technical vision systems. The aim of the invention is to improve the accuracy of determining the coordinates of the object. The device for detecting and determining the coordinates of an object in the image contains four memory blocks 1, 6, 8 and 9, a block 2 for calculating estimates of the brightness of image elements, a switch 3, a block 4 for estimating motion parameters, a block 5 for classifying objects, a block for differentiating 7, an address block 10 control unit 11 and block 12 coordinate registers. The purpose of the invention is achieved in that, depending on the mode of operation, a linear or non-linear estimation of the parameters of the object's movement trajectory, the classification of image elements into a plurality of background points and an interframe object filtering the background image elements and the object is performed. 11 il. sl C

Description

The invention relates to a technique of digital image processing and can be used in technical vision systems.

A device 1 is known comprising an analog-to-digital converter, three memory blocks, a correlation calculation unit, a valve, and a control unit. The drawback of the device is the low accuracy of determining the coordinates of the object, due to the presence of background areas, as well as additive noise in the reference image of the object.

A device 2 is known that contains an input memory block, a RAM block, a RAM block for storing correlation estimates, a correlator, an addressing block, a control block, in which the coordinates of the object in each frame are determined by finding the global minimum of the difference from two sets of correlation estimates. The first set is determined by comparing the reference with the actual current image, and the second by comparing the reference with the scaled current image, and the direction of scaling is determined by whether the object is approached or removed. The drawback of the device is the low accuracy of determining the coordinates of the object, due to the fact that the standard does not take into account changes in the shape and brightness of the current image of the object. In addition, along with the target object, there are background areas and additive interference that distort the reference image.

The closest in technical essence to the claimed device is

x | with XI

VI

cl

,

A device selected as a prototype for determining the position of an object in image 3, containing a first memory block, a temporal smoothing unit, a switch, a control unit, a second memory block, a correlator, and a coordinate determination unit. The first input of the first memory block is the information input of the device, the second input is connected to the first output of the control unit, the output of the first memory block is connected to the first input of the smoothing unit, to the first input of the switch and to the second input of the correlator. The second input of the smoothing block is connected to the second output of the control unit, and the output is connected to the third input of the switch, the second input of which is connected to the fifth output of the control unit. The switch output is connected to the first input of the second memory unit, the second input of which is connected to the third output of the control unit. The output of the second memory unit is connected to the first input of the correlator, the third input of the correlator is connected to the fourth output of the control unit, and the output to the first input of the coordinate determination unit, the second input of which is connected to the sixth output of the control unit. The first output of the unit for determining the coordinates is connected to the first input of the control unit, the second output of the unit for determining the coordinates is the information output of the device, the second input of the control unit is the synchronizing input of the device.

The drawback of the device is the low accuracy of determining the coordinates of the object in the image, due to the fact that a smoothed image is used as a reference image of the object, in which there are background areas along with the target object. In addition, the coordinates of the center of the reference image are used as the coordinates of the object, and not some point of the object, conventionally taken as its center. In this case, if the sample is updated frequently, the image of the object in the reference is shifted from the center to its edges, which leads to errors in determining its coordinates, and in the absence of a correction of the position of the object relative to the center of the standard, to the disruption of the tracking. A disadvantage of the device is the limitation of the functionality due to the lack of the ability to automatically detect small objects and track them.

The aim of the invention is to improve the accuracy of determining the coordinates of the object and expand the functionality of the device by solving

tasks of automatic detection of moving objects and ensuring the possibility of tracking small objects.

This is achieved in that the device

containing the first memory block, the first input of which is the first information input of the device, and the output is connected to the first inputs of the calculation unit

0 estimates of the brightness of the image elements and the switch, the second and third inputs of the switch are connected respectively to the output of the computing unit for evaluating the brightness of the image elements and the first output

5 of the control unit, the second output of which is connected to the first input of the second memory block, the second input of which is connected to the output of the switch, and the block of coordinate registers whose output is the output of the device, and the control input connected to the third output of the control unit, the fourth output which is connected to the second input of the unit for calculating the brightness of the image elements, and the input of the unit

5, the control input of the device is used to increase the accuracy of determining the coordinates of the object; a serially connected addressing unit is entered, the first input of which is connected to the fifth output of the control unit, the differentiation unit, the motion parameter estimation unit, the second input of which is connected to the output of the first a memory unit, and a classification unit, the second input of which is connected to the first output of the first memory unit, the second input of which is the second control input of the device, as well as the third and fourth blocks ki memory, wherein the sixth, seventh, eighth, ninth and tenth 0 th outputs of the control unit are respectively connected to the third input of the labeling unit, the second input parameter estimation block motion third input of differentiating unit, the first input

5 of the fourth memory block and the first input of the third memory block, the second input of which is connected to the switch output, the third input of the third memory block is connected to the third input of the second memory block, the fourth input of the classification unit, the first input of the first memory block and the second the output of the addressing unit, the third output of which is connected to the second input of the first memory block, and the second input is connected to

5 with the second input of the block of coordinate registers and the second output of the block for estimating motion parameters, the third input of which is connected to the third input of the differentiation unit and the first output of the fourth or fourth memory block, the second output of which

connected to the second input of the control unit, the third input of which is connected to the first output of the classification unit, the second output of which is connected to the third input of the image element brightness estimation unit, the fourth input of which is connected to the fifth output of the classification unit, the output of the second memory block and an input of the fourth memory unit, the third input of which is connected to the sixth input of the classification unit, the fifth input of the calculation unit for evaluating the brightness of image elements and the output of the third memory unit, and the fourth in od is connected to the first output of the addressing unit, whose third input is a control input of the third device, the fourth output block classification is the second output device.

FIG. 1 shows a block diagram of a device for detecting and determining the coordinates of an object in an image; in fig. 2 is a diagram of an addressing unit and a block of coordinate registers; in fig. 3 is a block diagram of the evaluation of the brightness of the image elements; in fig. 4 is a diagram of a differentiation unit; in fig. 5 is a control block diagram; in fig. 6 is a diagram of the first memory block; in fig. 7 is a block diagram of non-linear parameter estimation; in fig. 8 - classification scheme; in fig. 9 shows the timing diagrams of the signals explaining the operation of the device; in fig. 10.11 is a block diagram of the algorithm of the device.

The device for detecting and determining the coordinates of the object in the image (Fig. 1) contains the first memory block 1, the block 2 for calculating the estimates of the brightness of the image elements, the switch 3, the motion estimation parameters block 4, the object classification block 5, the second memory block 6, block 7 differentiation, a fourth memory block 8, a third memory block 9, an addressing block 10, a control block 11 and a block of 12 coordinate registers.

The addressing unit (Fig. 2) contains two address operators (FAs) 13, 14, four registers 15-18, four adders 19-22, three counters 23-25, a comparator 26 digital codes, two elements OR 27, 28, the element NOT 29.

The block of coordinate registers (Fig. 2) contains two registers 30, 31.

The unit for calculating the brightness of the image elements (Fig. 3) contains the element And 32, two registers 33, 34, multiplexer 35, subtractor 36, multiplier 37, adder 38.

The differentiation unit (Fig. 4) contains five registers 39-43, a subtractor

44, a RAM block (BOP) 45, a permanent memory block 46 (BPP), an element OR 47.

The control unit (Fig. 5) contains the shaper 48 micro-command addresses (FAM), the reprogrammable memory block 49 (PDU), the micro-command register 50, AND 51.52 elements, trigger 53, AND-NOT element 54, condition code multiplexer 55, counter 56, Element AND L AND-NOT 57.

The current image memory block 1 (Fig. 6) contains two information transceivers (PII) 59, 60, two BOP 61, 62, trigger 63, element 64,

5 Memory block 8 (FIG. 6) contains a BOP 65 and multiplexer 66.

The unit for estimating motion parameters (Fig. 7) contains registers 67-81, subtractors 82-84, three multipliers 85-87, four totalizers 88-91, two multiplexers 92, 93.

The classification block (Fig. 8) contains registers 94-100, two subtractors 101, 102, three multiplexers 103-105, four adders 106-109, four comparators 110-113

5 digital codes, BOP 114, And 115, 116 elements, counters 117,118, And 119,120 elements, two multipliers 121, 122, six PPB 123-128.

The device for detecting and determining the coordinates of the object in the image works as follows. The signal A1 (Fig. 9, c) opens the element And 52, to the input of which an external signal of the clock frequency A2 of the meander type (Fig. 9, a) arrives, specifying

5 frequency of operation of synchronized elements of the device. The frequency of the A2 signal is determined by the speed of the element base used. At the output of the element And 52, a signal A3 is formed (Fig. 9, b). At the same time, the unit signal level A1 is fed to the input R of the flip-flop 53 and the first input of the NE-HE element 54, to the second input of which an external frame sync pulse A4 is applied (Fig. 9, d). Unit level

5 of the next HR sync pulse sets the trigger 53 to a single state. As a result, at the output of the trigger 53 a signal A6 is formed (Fig. 9, d), which opens the element AND 51 and enables

0 the passage of control signals U1-U4 to the first control input of FAM 48. From this moment the device for detecting and determining the coordinates of the object in the image begins to work in accordance with

5 microprogram stored in reprogrammable memory block 49. The operation of the device is organized in such a way that during the time of one frame the current frame is recorded in one of the RAM blocks, for example, block 61,

block 1 of the current image memory and at the same time processing the previous frame, recorded in another block of the operating memory of block 1, is carried out.

To synchronize the recording of the current frame, external synchronizing signals are used: the frequency signal A7 of the square wave type (Fig. 9, h), which sets the sampling frequency of the samples of the input digital video signal A5, frame sync pulse A4 (Fig. 9, e), horizontal sync pulse A8 (Fig. 9, g) video signal and six-bit A14 signal. These signals control the counters 23 and 25, which form the ten-bit address signal A9 (Fig. 9, p), the OR element 27, which forms the zero write signal A10 A7VA4VA8VA12 (Fig. 9, k), the counter 24 and the digital comparator 26. The counter 24, to the counting input of which a horizontal sync pulse A8 arrives, and the input signal A15 (Fig. 9, m), and a digital comparator 26, comparing the digital values of the A13 signal (Fig. 9, p), taken from the output of the counter 24, and the signal A14, representing the thinning factor of lines n, form the signal A11 (FIG. 9, L), which appears with the output of the comparator 26 at the time of equality of the digital values of the signals A13 and A14. The signal permits the recording of each line of the image of a video signal sensor, a multiple of n, into memory blocks 61 or 62 of block 1. Signal A15 appears at the second output of digital comparator 26 in the case where the digital value of signal A13 is greater than the value of signal A14.

Counter 23, to the counting input of which the signal A10 arrives, and to the input signal set to zero, the signal A16 A12VA8, generates the addresses of the A5 signal samples in the row of the current frame (five lower-order bits of the A9 signal (A9,), and the counter25 controlled by inputs of the zero setting and the counting input with the A4 and A12 signals, forms the row address of the current frame (the five most significant bits of the A9 signal ().

The signals A9, A10 to the control inputs of the BNP 61 or 62 of the block 1 selected for recording the next television frame are transmitted by a four-channel transceiver 59 by switching channel B with channel A or C. The input of the selected BOP to the information input of the device is connected information transceiver 60 of block 1. The sequence of recording in BOP 60 and 61 of successively incoming frames of a television image is controlled by signals A17 (Fig. 9, i) and. A17 signals

and A18 are formed by a trigger 63, to the counting input of which an inverted frame sync pulse A4 is fed.

Channel X PPI 60 is connected to the information inputs of blocks 2-5. When the signal level A17 is zero, channel B of PPI 59 is switched to channel A, and channel X is connected to channel C, channel B of PPI 60 is switched to channel X, and channel C is connected to channel A. At unity

At the level of signal A17, everything is the other way round.

Channel A of the PIP 60 of block 1 is the information input of the device for detecting and determining the coordinates of the object at

5 image.

FIG. 9 shows time diagrams of signals, explaining the moment of the device start-up and the process of recording the video signal of the current frame, for example, in the BOP

0 61. The signal diagrams are qualitative, with the rows and the dimension of the current frame considered to be 8x8 elements. The bit diagrams,,, of the signal A9 in FIG.

5 9 are not shown because they have a zero value.

In the very first frame from the beginning of the operation, only the TV image is recorded, for example, in the BOP 61. The initial data arrays are formed and recorded in the corresponding blocks during the second frame recording in the BOP 62. The device for detecting and determining the object coordinates on the third frame image, in addition to recording the next frame in the BOP 61, handles the previous frame recorded in the BOP 62.

The algorithm for detecting and determining the coordinates of a moving object in the image, implemented in the inventive device, is obtained from solving the tracking problem as the problem of estimating motion parameters and parameters associated with the background and object images.

5 The block diagram of the algorithm implemented in the device for detecting and determining the coordinates of the object in the image is shown in FIG. 10, 11. The first part of the algorithm (Fig. 10) corresponds to the operation of the device in

0 in the mode of automatic detection and entry of the trajectory of the detected object.

FIG. 11 shows the second part of the algorithm of the device in the tracking mode.

5 In the algorithm (Figs. 10, 11) implemented in the claimed device, the following main information processing steps can be distinguished.

1. Formation and recording of initial data arrays G0, B0.

2. Classification of the elements of the observed image.

3. Recurrent filtering the motion parameters of the object and issuing coordinates in block 12.

4. Smoothing the reference image of the background Gn.

5. Smoothing of the reference image of the object.

6. Formation of the predicted image Sn + 1.

7. Calculation of the derivative image

neither

8. Nonlinear estimation of the velocity and coordinates of the object along the X axis.

9. Calculation of the derivative of the image of E5p nor-3m-.

o Any

10. Nonlinear estimation of velocity and object coordinates along the Y axis.

11. Classification of the trg region of the observed image.

12. Corrections of the object coordinates and their block 12.

The first computation step is performed during the recording of the second frame in the BOP 61 and only once. During this stage, the initial reference background image Go, the initial function B0 is formed and the contents of memory blocks 8, 9 and BOP 45 are zeroed. The initial reference image of the background Go is taken as the current image Lo recorded in the BEP 62. At that, There is no object in the image of C. The initial function B0 is formed by writing to the BOP 114 at all addresses of the number 32.

In the second stage of the calculation, for each element of the observed invention, Inij checks the condition

bnii b (n - i) ij + (Inij - Qnij) 2 (Inij

+

2 d

(D

where b (n-i) ij are elements of Bn-i; Inij - elements of the current image Ln; gnij - elements of the predicted image of the background Gn A1 - the predicted minimum difference of the object brightness from the background brightness,. As a result, if condition (1) is fulfilled for the element Inij, then the element is classified as an element belonging to the set Hn, otherwise as a background element. At the same time, the coordinates of the center of gravity and the area of the image of the object Hn are calculated. Coordinates of the center of tinplate

objects are calculated according to the expressions

. 2 i

3 InljeHn x Sn

I, I and InljeHn

  Sn

where Sn is the area of the object, equal to the number of elements that make up the image of the object Hn; i is the coordinate of the pc element

X axis; j is the Y coordinate of the Inij element. The second stage is performed in the classification unit 5. The classification of one Inij element is performed at the beat of the A3 signal cycle.

At the stage of classification, the inputs of the registers 96.97 receive, respectively, the values of the elements of the observed Inij (signal A20) and the smoothed reference background image gnij (signal A21) from memory blocks

1 and 6. The first input of the multiplexer 104 receives the value of the element of the smoothed reference image of the object hnij (signal A22) from the memory block 9. From the output of the register 97, the value gnij goes to the first

the inputs of the subtractor 101 and the adder 106. The second inputs of these elements receive the value A1 (signal A23), which is stored in register 94. At the output of the subtractor 101, the value (gnuj-Ai) (signal

A24), which is fed to the second address input of the PPO 123 and to the first input of the multiplexer 103. At the output of the adder 106, the value (gnij + AI) (signal A25) is generated, which is fed to the second address input

BPP 124 and the second input of the multiplexer 103. The first address inputs of BPP 123 and 124 are connected to the output of register 96. In PPO 123 and 124, respectively, the values of the modules of expressions C- (dcc-AI) and C- (dpu + AI) are written

for all possible combinations of C, (gnij- A i) and (Inij, (gnij + AI). The A26 signal generated by the comparator 110 controls the multiplexer 103. If I lnij- (gnij + Ai I llnij- - (gnij-A1) I , the signal A26 has zero

a level that connects the output of the adder 106 to the second input of the multiplexer 104. Otherwise, the output of the subtractor 101 is connected to the second input of the multiplexer. The multiplexer 104 is controlled

the most significant sign of the signal A27, which is the value of the element b (n-i) ij and is read from the BOP 114. If the element C before solving the classification problem belonged to the background, i.e. b (n-i), then to

the input of the register 95 is connected to the output of the multiplexer, 103., Otherwise, the output of the block 9 is connected to the input of the register 95.

On the second clock of the A3 signal, the value hnij (signal A28) is fed to the first address input of the PPO 125, and the second address input is fed to the value Inij. The first and second address inputs of the PPO 126 receive, respectively, Inij and gnij. In PPP 125 and 126, the values of the squares of the expressions (Inij-hnij) and (Inij-Qnij) are written, respectively. The outputs of the PPO 125 and 126 are connected respectively to the second and first inputs of the subtractor 102, the output of which is connected to the first address input of the PPO 127. The second address input of the PPO 127 receives a signal. In PPP 127, using the tabular method, the number arriving at the first address input is divided into 2D, where. As a result, at the output of the control unit 127, a signal A29 is generated, the value of which is ((Inij-gnij) - (Inij- -hnij) 2 / 2D. The value bnij is formed at the output of the adder 107.

In the third cycle of the signal A3, at the output of the multiplexer 105, an ASZ signal is generated, the value of which is equal to bnij, if -32 bnij 32; -32, if 32, if. The value of bnij (the ABZ signal) is recorded in the BOP 114 at address L (n-1) m. At the same time, the sign bit of the AZZ signal is fed to the input of the element HE 119 and the third input of the multiplexer 55 condition code. The signal is fed to the control input of the AND element 116, the counting input of the counter 117, and the fourth input of the multiplexer 55 of the control unit 11. The second input of the AND element 116 receives the value of the element address Inij (signal A69). If, then the coordinates are allowed (signal A69) through element 116 and the contents of counter 117 also increase. Five lower bits of the signal A69 are fed to the first input of the adder 108, and five high bits to the first input of the adder 109. Sum - the mattor 108 and the register 99 form an accumulating adder. The adder 109 and the register 100 form the second accumulating adder. At the output of registers 99 and 100, which are connected to the first inputs of multipliers 121,122, the value of 2 I (signal A37) and the value

lnijeHn

X j (A38 signal). In the counter 117 foriInij e NP

The area Sn of the set Hn is equal to the number of elements of this set. At the output of FOD 128, a signal A36 is generated, the value of which is 1 / Sn. The signal A36 arrives at the second inputs of the multipliers 121 and 122. As a result, at the output of the multipliers 121 and 122, the coordinates of the center of gravity of the object are formed on the X axis (A39 signal) and on the Y axis (A40 signal), which are fed to block 4. Counter output 117 connected to the first input of the digital comparator 113, the second input of which is connected to the output of the register of micro-instructions 50-3. A unit level signal A41 is generated at the output of the comparator 113 if the area of the object is greater than the number entering the second input of the comparator. The signal A41 is fed to the fifth input of the multiplexer 55 and to the first input of the AND 115 element. The output of the AND 115 element is connected to the counting input of the counter 118. The counter 118 is zeroed in with a unit signal level Y77 once in the first stage of the device operation. At the fifth bit of the output of the counter 118, a unit-level signal Y77 is generated at the moment when the trajectory of the object is closed, which corresponds to the detection of the object in 16 frames of the observed image.

In the third stage, the level of the A41 signal is analyzed in the control unit. If an object is detected at the classification stage, then the coordinates and velocities of the object are estimated in accordance with the expressions

APH Up + Kl (Anx ALH), (2)

 + K2 (Anx-Inx), (3)

Any Any + KI (Any Any), (4)

Any z Any-IO (Any - Any), (5)

AOX 0, Aoy 0, AOX 0, Aoy 0. (6)

Calculations of the estimates in block 4 for X and Y are performed sequentially and in a similar way, therefore it is sufficient to consider calculating estimates on the same axis, for example, X. A priori estimates of TOR and UPX are stored respectively in registers 75 and 77. In register 18, the value Any is stored, in register 76 - Apu. The first inputs of the subtractors 83, 84 are connected respectively to the outputs of the registers 70, 71, in which the calculated coordinates of the center of gravity of the TOR object (signal A39) and Any (signal A40) are written. The second input of the subtractor 83 - with the output of the register 77. Similarly, the second input of the adder 91 is connected to the output of the register 75.

At the output of the subtractor 83, a signal is generated - AH, which is fed to the second input of the multiplexer 92. At the output

multiplexer 92, an A46 signal is generated. The signal A46 is fed to the second inputs of the multipliers 88 and 87. The first inputs to the multiplier 86 are given a signal, the multiplier 87 is a signal. Ki and Ka are stored in registers 73, 74, respectively. At the output of multiplier 87, a signal is generated (Anx - Anx). The signal A49 arrives at the first input of the adder 90, the second input of which is currently connected to the output of the register 77. As a result, the output of the adder 90 forms the a posteriori estimate of the coordinate of the object Alx, which is written to the register 79, the output of which is connected to the first input of the multiplexer 93. The calculated the value of Apx through the multiplexer 93 is fed to the input of the register 80. At the output of the multiplier 86, a signal is generated (Aph - Alx), which is fed to the first input of the adder 89. The second input of the adder 98 receives the signal A53 APH- As a result The output of the adder 89 is the a posteriori estimate of the velocity of the object Ypx, which is written to the register 75. The a priori estimate of the coordinate on the (n + 1) -th frame I (n + 1) x is formed at the output of the adder 91 as the sum of the A53 / n signal and the signal and is written in the register 77.

l The difference in calculating the ratings Any and Any is that instead of register 77 to the inputs of adders 89, 91 register output 78 is connected, and instead of register 77 to the inputs of subtractors 83, 84, adder 90 connected register output 78, signal, and rating Any is written by a single signal level of A69 to register 81.

If the object in the current frame is not detected, then the signal has a zero level. In this case, the coordinates and velocity of the object are predicted in accordance with the expressions

(p-1) x + A (p-1) x, (7)

(n-i) x, (8)

Ah

Any A (n-i) y + A (n-i) y, (9)

(n-i) y (10)

and expression (6). This is accomplished by connecting during the third stage to the input of multipliers 86, 87 a signal A46 equal to the signal A45, which, in turn, has a zero value. The remaining calculations are similar to those described.

At the end of the third stage, the calculated estimates of the coordinates of the object's center of gravity, provided that the trajectory has occurred, are written to the registers 30, 31 of block 12, respectively.

At the fourth stage, in block 2, estimates of the values of the elements of the reference image of the background Gn are calculated in accordance with the expression

(eleven)

BUT

9nij gnij + K3 (| nirgnij),

where you are; d (p-odO KZ 1.

Reference Image Smoothing

BackgroundSp reduces the dispersion of additive noise present in the original image of the reference background, (2- Kz) / Kz times, as well as smoothing the sampling errors and quantization in various

frames. Limiting the Cs coefficient from below allows you to track changes in the background. At the fourth stage, the first input of multiplexer 35, which is connected to the output of memory block 6, receives the values

background elements cjnij (signal A21). At the first input of the subtractor 36, the values of the elements lnij (signal A20) of the observed image are received through the element AND 32. At the output of the subtractor, the signal A57 is formed, the value

which is equal to Inij-gnij. The signal A57 arrives at the first input of the multiplier 37. The second factor of the multiplier receives the short-circuit coefficient stored in register 33. The output of the multiplier is connected to the first input of the summator 38, the output of which produces the value gnij (signal A58). The calculated value gnij is written into the memory block ti 6. Similar calculations are performed only for elements of the observed image classified as background elements. The background membership check is performed by analyzing the sign of the value of bnij (AZZ-8 signal).

Reference Image Smoothing

The object Hn exists at the fifth stage in accordance with the expression

l

hnij hnij + K4 (lnij-rinij), (12)

. where (n-i) ij; .

The calculations performed in block 2 in the fifth stage are similar to those performed in the fourth stage. The difference is that in the register 33 the coefficient I0j is stored, and the second inputs of the subtractor 36 and the accumulator 38 are supplied with the value of the element hnij (signal A22), read from memory 9. The calculated value is recorded in block 9. Smoothing in accordance with expression (12) is performed

only for elements of Inij classified as object elements. The test for belonging to an object is carried out by analyzing the signal A34. Otherwise, the Y40 signal controlling the AND 32 element is zero. In this case, smoothing is performed in accordance with the expression

hnij hnij-k4hnij

(13)

This allows zeroing the elements of the memory area of block 9, which in the previous frames corresponded to the elements of the object, and at the current moment, due to the movement of the object, correspond to the background elements in the observed image.

At the sixth stage, in block 8, the predicted image Sn + 1 of the rectangular portion of the image Ln-H is formed, whose center coincides with the center of the object. The elements of the image are formed in accordance with the expression

Sn + Vf4 H (n + 1) ijJ1 h (n + 1) ifj1 + + (1 - r (n + l) i2J2) g (n + l) i2J2.

(14)

Where

Sn + elements of the plot of the predicted image Sn + 1 in the MxM dimension;

il V + inx-y, Jl / + Any-y,

lm

(n + l) x - 7pj2 iM + A (n + l) y

ML

- at; AH, Any - a posteriori estimates of the coordinates of the center of the same object in the observed n-frame, A (n + i) x, A (n + 1) y - predicted estimates of the coordinates of the center of the body of gravity of the object in (n + 1) - m frame; ir (n + i) iiji gpil 1, if, otherwise, r (n + 1) il - 0, for r (n + i) i2j2 t ° the same; h (n + 1) sludge

 PML 9 (n + l) i2J2 9ni2j2 Image formation Sn + 1 is carried out in two stages. At the first stage, the BOP 65 records the elements of the section of the image Np. Moreover, the value of h (n + 1) il (signal A21) and the signal A60 of the unit level are recorded in the BEP 65, if the signal A34 of the unit level, i.e. b (n + 1) sludge - 0 Otherwise, the BOP 65 records the signal A59, which has a zero value, and the signal A60, which has a zero level. As a result, an image is formed in the BOP 65.

S (n + l) v / r (n +1) ML n (n +1) I1J1

and binary image Z (n × -1) ift object masks, formed according to the rule

- ,,, l and .b (p-y) 1L 0 z (, b (n + i5 ,.

At the second stage, the values of the elements of the predicted background area are recorded in the BEP 65. And in BOP 65 is carried out

recording the value of the element g (n + 1) i2j2 (signal A22) in the event that the corresponding value of the element of the binary image Z (n + 1) Vfi (the signal A62, which is formed at the first output of the BNP 65 and arrives

on the sixth input of the multiplexer 55) is equal to

to zero. As a result, in the BOP 65 is formed

image S (n + i) v /, the values of the elements of which (signal A61) are read

from the second release of the BOP 65.

In the seventh stage, in block 7, the derivative of the predicted image Sn, formed in the previous frame, is calculated by the TOR parameter in accordance with

by expression

 (Sn V + 1C - Sn V - W)

 Zn Vfi,

where the value of Snv + 1 / g is formed at the output of register 39, Sn v-1 / g at the output of register 41, Znv / u. at the output of register 40. In the subtractor 44, the value of Sn v +1/4 - SnV-1 / g is calculated, which is fed to the first address input of the PPO 46. The second address input of the PPO 46 receives the ZnV / g. BPP 46 recorded the values of the expression

SnV + 1 "-SnV - -.

-2- -n

Vfi

 of all possible combinations (Snv-M // - SnV - i JnZnv / g). The computational value d SnVjU / dAnx is recorded in the BOP 45 at the address determined by the value of the signal A10, which, successively passing through the registers 42, 43, goes to the address input BOP 45.

At the eighth stage, in block 4, nonlinear estimation of the coordinate and velocity of the object along the X axis is performed in accordance with

where Sn vp are the elements of the predicted image stored in the BOP 65; d i

bnii b (n-l) ij +

, C pi - gmj) -CM - hnu Y 0

2D

elements of the derivative stored in the BOP 45; And v + Anx-y. ji / + Apuu; Ks,

KB - fixed coefficients of the covariance matrix of estimation errors Vn, selected from the range 0 (Ks, Ke) 1, based on the specified effective filter memory; x $ (n - 1) x + A (P -1) x, Any A (n 1) y + A (n - 1) y

During the eighth stage, the inputs of the registers 67, 68, 69 are received, respectively, by the values of the elements of the PC (signal A20),

, (signal A61),

dSnVft

--- 5-3-- (signal A66). o / n

The outputs of registers 67, 68 are connected together with the first and second inputs of subtractor 82, at the output of which the Imp value is formed (signal A67). The output of the subtractor 82 is connected to the first input of the multiplier 85, the second input of which is connected to the output of the register 69. In the multiplier 85, the value

dSnVU. „- 1 (. G N tz D (Inij - Sn vu), in this case

About Any.g

it is assumed that. The multiplier output is connected to the inlet adder input, which includes the adder 88 and the register 72. After the sum in register 72 is formed

M-1

s

(

35P3 | K

behind

ny

D (2nij-5n l,

the output of register 72 using multiplexer 92 is connected to the second inputs of multipliers 86.87. The first inputs of the multipliers 86, 87 are given respectively the values of the coefficients Ke, Ks, which are recorded in registers 73, 74. Further results related to obtaining the estimates Ap, Ap, A (P + 1) x D (p-M) x are similar calculations performed at the third stage in obtaining the corresponding estimates.

Calculations performed at the ninth and tenth stages are similar to calculations performed at the device at the seventh and eighth stages. The difference is that at the ninth stage the elements of the image Sn v / g and 2. п v / u are read from block 8 not by rows, but by columns.

At the eleventh stage, the ifa section is classified by the size MxM of the observed image Ln by checking for each element Inij the conditions

(17)

0

five

0

five

0

five

0

five

0

five

L Center strobe trp has coordinates Alh. Any, which are defined in the eighth and tenth stages. As a result of this step, the elements of the area are divided into a plurality of background points and a plurality of object points. Simultaneously with the classification, the area and coordinates of the center of gravity of the selected set of object points are determined. The calculations performed in block 5 in the eleventh stage are similar to those performed in the second stage.

At the twelfth stage, the corrections of Lpx and Llu estimates found at the eighth and tenth stages are carried out by replacing them with the coordinates of the center of gravity of the object obtained at the eleventh stage. This is done as follows. The coordinates of the center of gravity of the object — the signals A39 and A40 — through multiplexer 93 are alternately recorded in registers 80, 81. At the same time, the predicted coordinates of the object A (n + i) x D (n-H) y stored in registers 77 are corrected. 78. After this, the contents of the registers 80, 81 are overwritten into registers 31, 30 of the block of coordinate registers.

An important role in the implementation of these stages of the algorithm belongs to the block addressing. The addressing unit is designed to form addresses of the elements of the current image, smoothed reference images by the object backgrounds, the predicted image Sn and its derivative with respect to the parameter and the classification function Bp.

The formation of the addresses of the elements of the predicted image Sn and its derivative with respect to the parameter Yapp or Any is performed using FA 13.14 and register 17.

The addresses of the elements of the current image being processed, the smoothed reference images of the background and the object, and the classification function are generated using the FAs 13, 14, adders 19-22, registers 16-18.

The operation of the device for detecting and determining the coordinates of an object in the image is controlled by a control unit, which is a firmware control unit. Each of the listed processing steps of the current frame corresponds to a specific micro-command or group of micro-commands that make up the microprogram of the implementation of this stage. All control signals (Y1-Y80) are recorded in microprogrammable memory 49. Block 49 is a reprogrammable memory block. When the corresponding address is fed to the address input of the BRP49, the selected values of the control signals are read, which are then written to the microcommand register 50 by the signal A3. The microcommand address is generated by the microcommand address driver 48.

The microcommand formed by the control unit has the following format. The signals Y1-Y18 form a microcommand field, reserved for controlling the generation of the address of the next microcommand, the signals Y19-T26 constitute the second field of the microcommand used as a data field, which are required in the processing of information, the signals Y27-Y80 constitute the third field of the microcommand and control the operating part of the device.

The synchronization of the control unit, which controls the execution of the processing steps of the next frame and their sequence, with the recording of the input signal A5, is achieved by fixing the end time of the next frame sync pulse A4 by the microproject address generator 46. Such a moment is the transition of the signal value from one to zero. For this, the signal A4 and the signal A74 are supplied respectively to the ninth and tenth input of the multiplexer 55 condition code. The determination of the end of the frame sync pulse A4 executes the microprogram of fixing the moment of the start of recording a new frame, consisting of two microcommands. Under the control of the first microcommand, a single level of the A4 signal is fixed. If the A4 signal has a single level, then the microinstruction is executed during one period of the clock signal A3, if not, then the microinstruction will be repeated until the A4 signal receives a single value, then the control is transmitted to the second microiname, identical to the instruction ( Y1-Y4) to the first microcommand, but fixing the zero level of the A4 signal. After the second micro-command has been executed, which determines the end of the A4 frame sync pulse and the start of recording the next new current frame, the device control in accordance with the instruction stored in the second micro-command field (Y1-Y4) is transmitted to the first micro-command of the first microprogram. stage.

The device allows to increase the accuracy of determining the coordinates of an object by allocating a plurality of image points of the object, using the calculated coordinates of the object's center of gravity as the object coordinates and periodically correcting the image position of the object in the reference relative to the center of the reference. If the complex of these measures is not performed, then the interframe smoothing of the reference image of the object displaces the image of the object from the center of the reference to its edges. This, when used as the coordinates of the object, coordinates of the center of the reference, an error occurs in determining the location of the object, which in the absence of correction over time accumulates and leads to a breakdown of tracking.

The device allows you to increase the accuracy of determining the coordinates of an object by applying the trajectory of processing the measurement results, which, firstly, allows smoothing and thereby reducing the errors that occur when measuring the coordinates of an object, and, secondly, by predicting the position of the object in The following frames allow not to lose the controlled object in case of its temporary disappearance, for example, due to its closure by an unauthorized object or accidental failures.

In the device, at the stage of nonlinear estimation of object coordinates, elements of the predicted (reference) image belonging to the set of object points are taken into account. Elements belonging to the background are not counted. It also improves the accuracy of determining the coordinates of the object.

The device has a wide functionality. In addition to tracking extended objects, the device performs automatic detection of small and extended objects appearing in the observation area, as well as tracking small objects with automatic transition from the detection mode to the tracking mode.

Claims (1)

The invention The device for detecting and determining the coordinates of an object in the image contains the first memory block, the first input of which is the first information input of the device detecting and determining the coordinates of the object on the image, and the output is connected to the first inputs of the unit for calculating the brightness of image elements and the switch, the second and the third inputs of the switch are connected respectively to the output of the unit for calculating the brightness of the image elements and the first output of the control unit, the second output It is connected to the first input of the second memory block, the second input of which is connected to the switch output, and the block of coordinate registers, the output of which is the first output of the object detecting and determining coordinates of the object in the image, and the control input connected to the third output the control unit, the fourth output of which is connected to the second input of the evaluation unit of the brightness of the image elements, and the input of the control unit is the control input of the device for detecting and determining the coordinates of the object in the image, characterized in that, in order to improve the accuracy of determining the coordinates of an object, a serially connected addressing unit is introduced, the first input of which is connected to the fifth output of the control unit, the differentiation unit, the motion parameter estimation unit, the second input of which is connected to the output of the first memory block and the object classification unit, the second input of which is connected to the first output of the first memory block, the second input of which is the second control input of the device for detecting and determining the coordinates of the object on The third and fourth memory blocks, while the sixth, seventh, eighth, ninth and tenth outputs of the control unit are connected respectively to the third input of the object classification unit, the second input of the motion parameter estimation unit, the third input of the differentiation unit, the first input of the fourth memory block and the first input of the third memory block, the second input of which is connected to the switch output, the third input of the third memory block is connected to the third input of the second memory block, the fourth input of the classification block of the objects, the first input of the first memory block and the second output of the addressing block, the third output of which is connected to the second input of the first memory block, and the second input is connected to the second input of the coordinate register block and the second output of the motion parameters estimator, the third input of which is connected to the third the input of the differentiation unit and the first output of the fourth memory unit, the second output of which is connected to the second input of the control unit, the third input of which is connected to the first output of the classification unit, the second output of which connects It is connected to the third input of the image element brightness estimation unit, the fourth input of which is connected to the fifth input of the classification unit, the output of the second memory block and the second input of the fourth memory block, the third input of which is connected to the sixth input of the classification unit, the fifth input of the calculation unit estimates of the brightness of the image elements and the output of the third memory block, and the fourth input is connected to the first output of the addressing unit, the third input of which is the third control input of the detection and determination device coordinates of the object in the image, the second output of which is the fourth output of the classification unit. figure 1 LO Yu GGE r8: U19-U26 Ultrasonic 33 A3, A3 A62 A61  A61 ilTl Y42 A3   | A21 A22 35 Y37 36 37 34f / 39 Fig.z It A70 Ml 73 FIG A yg-f / fs A9 MO A69 FIG. five A3 U19-U26 Y50 U5b U55U56U60 U61 A6s AJ9 Phage. 7 C, AM $ A56 113 U80- g 8 6Ш 11 AW FIG. 8 Lerba string First line about current u. Frame Eighth row dte cusch. and // 9/9 / 9V) / Y T / TJ /// 7 - -. . AT . «Ъ ч JTW« «« .and j, s Q / fft l UH /  Start Nl / Li Formation ref. reference background, classification functions, etc. GT m and fixing the start of the next frame Sh L Classification is evil. current frame gu - UbcKft / Stream Frame - Detected / Yes 51 Calculated, estimated. coordinates the object ппхЛпии прог--. nozap 1x, l / t / s / 13 W / fQ ZakL trajectories - Lchalis1. Yes About Issuance In block 12 lph IT Ш Smoothing of the balm is depicted. Background plet. ShValizirovanie reference images. an object AND Not wore K block P i Yes Not 3L SJ Calculation of prognosis, n + iy pg W issuing b 5lok 12 forecast nx, pu g / Fig.Yu L to block 2 1ZJ Formation of the predicted image Sn + j gu Jz. Jz. JS1 Fixing the start of the next frame BUT. Calculation Schneline estimation of coordinates fast, lph object calculation ML l „„ 1 Calculate L ELPU 22J Nonlinear estimation of the coordinate and speed, llu object Calculation s Classification plot current. frames with Yal / Dlurp center coordinates th / j Correction of GLH, LPU, taking into account the Vychisl coordinate. center of the object LH, dts Qg / 10 / f 2PG 5j Slajibanis standard foot background image pg Ji   - W Smooth & a reference image of the object background Ji pg -