JPH01281579A - Image processor - Google Patents

Abstract

PURPOSE:To simplify and miniaturize the constitution of the title processor and to rapidly execute calculation by serially connecting digital differential analytical operation circuits having the same constitution, extracting the feature point of an original image and computing a rotating motion recurrence formula to omit the operation of an initial value. CONSTITUTION:The edge of a picture element signal of an original image inputted by a camera 11 is detected and inputted to a preprocessing part 17 as edged data. The preprocessing part 17 executes prescribed preprocessing and sends the processed data are sent to the digital differential analytical(DDA) operation circuits 18. Then, the operation of the rotating motion recurrence formula for finding out a Hough curve is executed by the circuits 18. Said operation is continued until all processing of picture elements other than a window or picture elements more than a threshold out of picture element signals on the surface of the original picture to be processed. After ending the processing, the filtering processing and sorting processing of the intersecting point of Hough curves are executed by a near-by filter 19 and a sorting part 20. An approximate line of a curve connecting points to be processed on the original image can be found out as a final result.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to an image processing device, which is used for controlling unmanned vehicles, real-time recognition of the motion of an observed object, and the like.

[Conventional technology]

For example, when controlling unmanned robots or autonomous vehicles, it is necessary to use a camera to capture images of identification lines displayed in advance on the road ahead, or the center line or shoulder line of the road, and process the images in real time.

Figure 15 is for explaining road recognition using images, where (a) is an image captured by a camera, and (b) is an image captured by a camera.
is the luminance (or its rate of change, etc.) in figure (a)
3 is a diagram showing pixels with high values as black dots. FIG.

As shown in FIG. 3(a), the camera image 1 shows a road 3 extending toward infinity from the horizon 2. Roadside lines 4 are drawn on both sides of the road 30, and a center line 5 is drawn in the center. is depicted. Here, the shoulder line 4 and center line 5 of the road 3 have higher brightness than other parts,
Therefore, in the same figure (b), dots 4' and 5' are placed in these parts.
will appear continuously. Camera image 1 like this
In order to recognize the traveling direction and curved shape of the road 3, an approximate straight line of the curve connecting dots 4', L2. It is sufficient to recognize L3 etc.

Traditionally, as a method for finding such an approximate straight line,
A technique called Hough transform is known (for example, US Pat. No. 3,069,654). This is the first
This will be explained with reference to FIGS. 6 to 18. As shown in FIG. 16(a), when a processing target point P (x, y) exists in the original image represented by the x-y coordinate system, a p straight line lCf1. I, etc.) can be drawn infinitely.

Then, it intersects perpendicularly to this straight line N, N,... and the origin is 0
Regarding the straight line passing through (0,0), the straight line It, i
),... You can draw one line for each.

b Here, regarding the straight line passing through the origin 0 (0,0), direct t
*1) The length up to (N, 11., etc.) is ρ (ρ .

a
aρ, etc.), and the angle between it and the X axis is θ (θ , θ.

b
a), then the above ρ of the straight line passing through this origin.

θ is expressed as a sine curve, ie, a Hough curve, as shown in FIG.

Here, the origin 0 (0, O) and the processing target point P (x
, y) is the distance ρ of this processing target point 1)
I) The longest among the above ρ (ρ, ρ6...) regarding IaX is 21/2 ρ13, -(Xp2+y), and when θ-0, it becomes ρo-+-z.

Next, three points P lined up in a straight line as shown in Figure 17(a)
When the Hough transformation shown in Fig. 16 is applied to ~P3, the sine curve (Hough curve) for point P1 becomes as shown in the dotted line in Fig. 17(b), and the sine curve for point P2 becomes as shown in Fig. 17(b). The sine curve for point P3 becomes like the dashed-dotted line in FIG. Here, the peaks (ρ , θ ), (ρ , θ ) and (ρ , θ ) of the sine curve in FIG.
θ) corresponds to 3, which corresponds to the distances ρ1 to ρ between the origin O(0, O) and the points P, P, and P, and the angles θ to θ3 with the X axis, respectively.

In Figure 17(b), if we focus on the point where the three Hough curves (sine curves) intersect, we can see that the coordinates here are (
ρ , θ, ), which is the same as the ρ , θ of the straight line passing through the origin 0 (0,0), which is perpendicular to the straight line in Figure (a).
is equal to Therefore, if we find the intersection point of such a sine curve, we can find the approximate straight line of the curve between the dots (black dots) drawn in the x-y orthogonal coordinate system of the original image (however, in Fig. 17, this curve and the approximate straight line are ) can be found.

To explain this with reference to Fig. 18, first, in Fig. 18 (a), there are many dots (processing target points) to be Hough-transformed on the x-y coordinate plane (original image plane), and these are lined up on a curve. shall be.

Here, in the same figure (a), there are three approximate straight lines in the curve connecting the dots, L2. Can draw L3.

Therefore, if all of these dots are converted into sine curves (Hough conversion) as shown in Figure 16, we get
Intersection points of the sine curve as shown in FIG. 16(b) are obtained centered on three points. The coordinates of this intersection point are shown in FIG. 18(a) (ρ θ ) and (ρ1°).

tt' tt θ ) and (ρ θ ), so this t2
If 13' t3 is expressed in the coordinate system of ρ, θ, H, where H is the frequency of appearance of the intersection point, the result will be as shown in FIG. 3(b).

Therefore, the approximate straight lines L1 to L3 of the curve corresponding to the shoulder line 4 of the road 3 as shown in FIG.
This can be determined by the values of ρ and θ at the peak of H (frequency of appearance of crossover points) in Figure (b).

[Problem to be solved by the invention]

However, it has not been easy to apply the Hough transformation method as described above to high-speed and real-time image processing. This is because the coordinate values (x,
When trying to find the distance from the origin to the straight line g in order to find the p Hough curve (sine curve) based on
・slnθ+y ``cosθ ・(1)p must be executed.For example, if θ is divided into 512, trigonometric function calculations must be performed 1024 times, multiplications must be performed 1024 times, and addition must be performed 512 times.And, If the original image to be calculated is composed of, for example, 512×512 pixels, the total number of calculations will be extremely large, and processing with a normal processor will significantly increase the processing time.

Of course, the sin θ required in the above equation (1)
It is also possible to shorten the time required for calculation by storing the values of F.R.
"Hough conversion hardware using OM" 1986 IEICE 70th Anniversary General Conference, No.
, 1587). However, this method requires a large-capacity ROM, and the access time cannot be ignored. Furthermore, in this way, data of trigonometric functions can be stored in ROM.
Even if it is, the number of multiplications in the calculation of equation (1) remains the same as before, and considering that the multiplication time is considerably longer than the addition time, this cannot be a fundamental solution.

On the other hand, if we consider the above-mentioned document "Hough conversion hardware using FROM" from a different perspective, here we will find the above-mentioned (
1) By parallelizing the calculation unit that calculates the expression,
The aim is to speed up overall signal processing. However, in this case, the number of memory tables (ROMs) for determining sin θ and cos θ will be equal to the number of calculation units connected in parallel, resulting in an extremely large-scale hardware system.

For this reason, it has been almost impossible to control a vehicle traveling at high speed in real time based on image data, or to recognize an observed object moving at high speed in real time based on image data. Furthermore, when attempting to increase the speed by parallelizing the arithmetic units, it was inevitable that the hardware would become larger.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an image processing apparatus that is not only capable of processing image data at high speed in real time, but also capable of reducing the size of the hardware.

[Means to solve the problem]

An image processing apparatus according to the present invention is an image processing apparatus that separates and extracts feature points of an original image from a plurality of processing target points on the original image captured by an imaging means, and includes the following elements. That is, DDA (Dlg
ltal DIHerentlal Analysls
:Digital differential analysis) DDA calculation means configured by connecting a plurality of calculation elements in series; storage means for storing calculation results of the DDA calculation means in correspondence with the DDA calculation elements;
The present invention is characterized by comprising means for extracting feature points of the original image based on the contents stored in the storage means.

Further, an image processing apparatus according to the present invention is an image processing apparatus that derives an approximate straight line of a curve connecting a plurality of processing target points on an original image captured by an imaging means, and includes the following elements. In other words, let the coordinates of one point on the circumference of an approximate circle drawn in the α-β orthogonal coordinate system be (α, β1), and let the rotation angle to the coordinates (α, β) of the next point on the circumference be ε. Then, for 1+1 1+1 (where i is a positive integer), the rotational motion recurrence formula is α, +1−f (α, βI, ε) α 1 β −f (α, β1.ε) Among the DDA calculation means that sequentially calculates each predetermined rotation angle using the Vibration method, and the results sequentially calculated by this DDA calculation means,
It is characterized by comprising a storage means for storing at least the value of β1, and an approximate straight line deriving means for deriving an approximate straight line from the intersection of Hough curves for each of a plurality of processing target points obtained based on the stored contents of the storage means. shall be.

[Effect]

According to the configuration of the present invention, a calculation unit can be easily formed by simply connecting DDA calculation circuits (calculation elements) of the same configuration in series, and there is no need to refer to a memory table or the like during calculation. It becomes significantly simpler and smaller.

Further, by configuring the arithmetic circuit to execute the rotational motion recurrence formula in a pipeline manner, it is possible to speed up calculation.

[Example of real power]

Below, based on Figures 1 to 14 of the attached drawings,
The present invention will be described in detail. In addition, in the description of the drawings, the same elements are given the same reference numerals, and redundant description will be omitted.

FIG. 1 is a block diagram showing the overall configuration of an image processing apparatus according to an embodiment. In the figure, a camera 11 captures an original image by capturing an object to be processed (for example, a road, a high-speed moving object, etc.), and this image signal is digitized by a signal input section 12 and sent to an edge detection section 13. . The edge detection unit 13 extracts the edges of the image signal and generates edge data with shading, for example, 512
The signal is sent to the multilevel memory 14 as ×512 pixel signals (edge pixel signals). The multilevel memory 14 stores edged data for each pixel, and each time the scanning of one screen is completed, the edged data is sent to the D/A converter 15 and given to the CRT display 16 as an analog signal.

Therefore, this edge data is displayed on the CRT display 16.
is displayed.

On the other hand, the edged pixel signal output from the edge detection section 13 is given to a preprocessing FJ section 17, and the edged pixel signal subjected to preprocessing as will be described in detail later is given to a DDA calculation section 18. The DDA calculation section 18 includes n DDA calculation circuits 18 to 18, which are connected in series.

A neighborhood filter 1 is provided on the output side of the DDA calculation unit 18.
A sorting section 20 is connected to one of the two, thereby performing neighborhood filtering processing and sorting processing (described in detail later).

Note that the above circuit elements are connected to the CPU via the VME bus 21.
22 to control signal processing operations and synchronize processing timing. Further, the preprocessing section 17, the DDA calculation section 18, and the neighborhood filter 19 are connected to each other via the VME bus 23, and synchronous control of the transfer of the DDA calculation results and the transfer of the gray value data is performed.

Next, the detailed configuration of the main parts of the image processing apparatus shown in FIG. 1 will be explained.

FIG. 2 is a configuration diagram of the preprocessing section 17 and DD in FIG.
It corresponds to the A calculation section 18 and the neighborhood filter 19.

As shown in the figure, the preprocessing unit 17
t-'In First-Out) board 17'. The FIFO 17' receives the coordinate values (X, Y) of the point to be processed on the X-Y plane as an address p signal, and also receives the edged gray value data DI as a data signal. As will be described later, this FIFO 17' performs coordinate conversion from X-Y coordinates to x-y coordinates, window setting, and threshold processing, and sequentially outputs the results according to the FIFO method.

Each stage of the DDA operation circuits 18o to 18 shown in FIG.
．． 33, and F/F 31 receives address signals α to α
, β to β are temporarily stored On-1On-1 respectively, the F/F 32 temporarily stores the gray value data D1, and the F/F 33 is stored in the RAM 34 (R
Histogram data DMo to DM(.

-1) is stored temporally. DDA37 (DDA-D
DA) calculates the rotational motion gradual n-1 formation described later for each rotation angle, and inputs the address signals α, β1 and outputs the address signals α1+1, . β
Output. The adder ADD1+135 (ADD ~ ADD) is FIFO17'
The gray value data DI from On-1 and the histogram data DMo-DM(.-1) from the RAM 34 are added, and the output is temporarily stored in the buffer 36 and then stored in the RAM-RAM. is sent to each of n-1. The timing controller 25 controls the timing of signal processing in these circuit elements, and generates a timing pulse φ.

Output ~φ. It is connected to a command/status Q interface (1/F) not shown.

Next, the overall operation of the image processing apparatus shown in FIGS. 1 and 2 will be explained with reference to FIG. 3.

FIG. 3 is a flowchart explaining this.

First, a pixel signal for each point to be processed on the original image captured by the camera 11 is inputted via the signal input section 12 (Step 102), and edge detection is performed by the edge detection section 13 (Step 104) to convert the edged data. The preprocessing section 17 is manually inputted (step 106). Above steps 102-1
The process of step 06 is repeated each time the pixel signal is manually input, and the result (edged data) is sequentially sent to the signal preprocessing unit 17.
is sent as digital data.

The preprocessing unit 17 performs predetermined (described later) preprocessing (step 10
8) and sends the processed data to the DDA calculation unit 18.
(step 110). This preprocessing is also sequentially repeated each time edged data is provided.

Next, the calculation of the rotational motion recurrence formula for obtaining the Hough curve (sine curve) is executed as <DDA calculation (step 112), as will be explained later. Of the pixel signals of one screen (original image plane) to be processed, processing is continued until the processing of all pixels other than those excluded by the preprocessing unit 17 as outside the window or below the threshold value is completed (step 114),
Once completed, filtering processing (step 116) and sorting processing (118), which will be described later, are performed on the intersection of the Hough curves by the neighborhood filter 19 and the sorting unit 2.
0, and as a final result, an approximate straight line of the curve connecting the points to be processed on the original image is obtained.

Next, the edge detection method and edged data in the edge detection section 13 will be explained with reference to FIG.

The original image captured by the camera 11 is shown in Figure 4 (a).
If we take out the line indicated by reference numeral 8 in the figure using the X' coordinate, we can see that the luminance S is shown in analog form as shown in figure (b). That is, road 3
The brightness of the outer part of the road 3 and the road shoulder is low, but the brightness of the road shoulder line 4 is very high. Here, in Figure 4 (
In the embodiment, the brightness distribution as shown in b) is recognized as digital data of 256 gradations, but understanding the distribution of brightness itself is not sufficient to accurately recognize the shape of the road. .

Therefore, if the luminance S of the pixel signal obtained through the signal input section 12 is differentiated with respect to the coordinate X'(ds/dx') and understood as the rate of change in luminance, the edge will appear as shown in Fig. 4(c). This becomes clear, and when expressed as an absolute value lds/dx'l, it becomes as shown in FIG. It is obtained by This edge detection section 1
The edged data from 3 is subjected to predetermined preprocessing in the preprocessing section 17 in the next step.

FIG. 5 is a flowchart for explaining the preprocessing in the preprocessing section 17 (PIFO 17').

First, the edge data obtained as shown in FIG. 4 is input from the edge detection section 13 to the preprocessing section 17 (step 122), and it is determined whether or not this data is within a preset window. (Step 124). This window is set, for example, as shown by reference numeral 9 in FIG. 4(a). Here, whether or not the edged data is inside window 9 can be determined by the coordinate value of the data on the X-Y coordinate plane, and the next step 126 is executed only for the edged data inside window 9. Ru.

In step 126, whether or not the edged data in the window 9 is equal to or higher than a predetermined threshold value is digitally determined using, for example, a look-up table (LUT) attached to the PIFO 17'. That is, as shown in FIG. 6, when no LUT is provided, the input data and output data of PIF017' correspond one-to-one (as shown in FIG. 6(a)); Take the level for example! When set to th,
When the human power data is less than ■th, the output data becomes 0 (zero). Note that when this edge data is understood in 256 gradations, the threshold level 'th is θ~2
It can be set arbitrarily between 55 and 55. If this threshold value is shown in analog form, it is set as shown by the dotted line in FIG. The data (digital data) corresponds to 5. Therefore, only the main data (for example, data corresponding to the shoulder line and center line of the road) need to be subjected to signal processing, which will be described later, so it is not affected by noise components and the overall processing speed is reduced. It can be made faster.

Next, in step 128, a coordinate transformation is performed.

That is, the X-Y coordinate shown in FIG. 4(a) is converted into the X-y coordinate. As a result, the coordinates (X, Y) of the processing target point in the original image are converted to coordinates (x, y) in the Xp-y coordinate system. The above processing completes the preprocessing for Hough transformation.

Note that the order of steps 124 to 128 in FIG. 5 may be different. For example, the coordinate transformation in step 128 may be performed first, but in consideration of the time required for data processing, it is considered most preferable to perform the coordinate transformation in the order shown in FIG.

Next, application of the Hough transform in this embodiment will be specifically explained with reference to FIGS. 7 and 8.

As already explained in Fig. 11, if a Hough curve (sine curve) is obtained for the point P (x, y) shown in Fig. 7(a), it will be as shown in Fig. 7(C). be. By the way, it is also easily understood from the theorem of trigonometric functions that the locus of such a sine curve can be replaced with the locus of circular motion as shown in FIG. In other words, Hough about point P (x, y) in figure (a)
Executing the transformation p to obtain the Hough curve shown in Cc) in the figure is equivalent to finding the circumferential locus of circular motion as shown in Fig. 2B. Here, the radius R of the circle in the same figure (b) is 22.1/2...(2) R-ρ - (x + y 118 Starting from p, its initial value θ6 is θd-π/2-θ□8 However, tan θ-x/y... (3) I
aX I) p.

The inventor focused on this fact and applied the acclimatization formula for circular motion when drawing the circle in FIG. 7(b),
We have found a method that can easily perform Hough transformation of point P (x, y) to p (c) in the same figure. Here, according to the recurrence formula for circular motion above, α−
The coordinates (α , β
), when i is a positive integer and 1+1 1+1, al or l-f (α, β, ε) α 11 β −f (α, β, ε) ...(4) Country
It is determined as β jl.

Several specific contents of this equation (4) have been known for a long time. For example, the rotation angle ε is expressed as ε-2-"(r
ad) (However, when m-0, 1, 2, -), α −α −2βi Country 1 β −2α1+βI ... (5) 1+
1 or al・1″″a1-2 βI β −2α tenβ, (1-2)...(6) Country
1. In addition, the present inventor found that the calculation is more accurate and easier: 2m-1 α ■α (1-2)-2β1 1+1 1 -2m-1> β -2α +β, (1-2 1+1 1... (7) Or -1-3 intestine+β (-2+ε/6) 2g-1 +β (1-2)...(8) etc. may be used.

Therefore, if we apply the recurrence formula of equation (7) above, the second
Prior to explaining the specific operation of the circuit shown in the figure, this calculation method will be specifically explained.

FIG. 8 is a flow chart thereof. First, data preprocessed by PIFO 17' according to FIG.
In step 132), the processing target point P(x) is input to the DA calculation unit 18 in FIG.

y), i.e., ρ −X and p

Opening position (angle) θ' (-θo) = O(r
Start the circular motion from ad). At this time, FIFO
(α, β) from 17' is α −y 1βo−1z, ...(9)
p. Here, by setting the starting position of the circular motion as described above, an initial value for calculating the recurrence formula becomes substantially unnecessary.

Next, after storing the above β value in RAMo, (7
) to find α゛ and β1. ■ Substituting α and β obtained from equation (9) into equation (7), D
can be determined from the output of DAo (step 134)
, DDA, DDA, DDA3, . . . , the results (β, β2, β3, . . . ) are sequentially stored in RAM, RAM2, RAM3, . . . (step 138).

On the other hand, between step 134 and step 138, the gray value data is accumulated. That is, RAM34
Histogram data D read from (RAM, )
The gray value data from M1 and FIFO 17' are added and this is stored in the RAM again (step 136).
.

Then, step 140) repeats this calculation for each rotation angle ε until it goes around the circle.
The Hough curves for the two processing points are the values of β0゜β , β , β , ... stored above and θ
, θ , θ2°123 at .

No Ml...DM(.-1)) is also determined. Hereinafter, when the processing shown in FIG. 8 is executed for all processing target points on the original image, a plurality of Houg
The h curves will be determined in the ρ-θ coordinate system, and these will have intersection points as shown in FIG. 17(b).

Next, the operation shown in the flowchart of FIG. 9 will be explained in more detail with reference to FIG.

First, data input in step 132 in FIG. 8 is performed by inputting a ready signal from the FIFO 17' in FIG.
After inputting the coordinate values (x, y) of the processing target point P
P address signals α and β0 corresponding to FIFO 17' to F/F3
1, and input the grayscale value data DI of the processing point P to F.
This is done by inputting /F32. Here, the address and data input to this F/F 31.32 are the timing pulse φ from the timing controller 25.
It is done in synchronization with. And this timing pulse φ
Address signals α and β of the F/F 31 are inputted to the first DDAo (37) in synchronization with the rising or falling edge of 3.

In this DDAo, the process of step 134 in FIG. 8 is performed. That is, the calculation of the recurrence formula according to the above-mentioned equation (7) is executed, and the calculation results (address signals α 1 , β 2 ) are sent to the next DDAl (not shown). Here, the above recurrence formula (7) basically does not include trigonometric function calculations or multiplications, and the memory table (
Since there is no need to refer to the ROM, calculations can be performed easily and quickly. These have sufficient accuracy (few errors) to perform circular motion.

Address signal β stored in F/F31. is RAM, (
34), and thereby the histogram data D stored in RAMo is read out. That is, RAM O has the rotation angle θ with O β as the address. Histogram data D regarding the processing target point of the pond corresponding to (θ' −0)
is stored in advance (by the preceding ADDo operation), and therefore, by giving the address signal β from the F/F 31 to the RAMo, the histogram data is transferred from the RAMo to the F/F 33 in synchronization with the timing pulse φ. DMo will be sent.

Next, ADD (35
), but this A
The density value data DI of the processing target point (x, y) that is the calculation target is given to DDO from the F/F 32. Therefore, A.D.D.
Now let's look at the rotation angle θ and address β that had been stored in RAMo up until then. Histogram data DMo corresponding to
and the rotation angle θ of the processing target point (x, y) during processing. and gradation value data DI corresponding to attopresβ are added (step 136 in FIG. 8). Then, this addition result (D
' -DMo+D, ) is temporarily held in the buffer 36 and then sent to RAMo in synchronization with the timing pulse φd, and the address β0 of RAMo corresponding to θ is stored according to step 138 in FIG. It will be done to

The above one cycle processing is carried out by the DDA arithmetic circuit 18,
(L -1, 2, -n -1) is explained as shown in Fig. 9.

First, for the coordinate values (x, y) of the point to be processed, p The corresponding address signals (initial values) α and β are α.

β → α , β → α , β →...α , βI and 0
1122 + Address signal α1゜β as a result of sequential calculation and the gray value I of this processing target point (x, y)
pp data DI is input from F/F 31 and 32 and held (step 202), and address signal α is input from F/F 31.
, β are sent to DDA, IZI, the address signal α1 . β
The calculation of recurrence formula based on DDA.

It is executed in Then, the resulting address signals α and β are 1+1 of one cycle of processing.
1+1 After completion, the next DDA calculation circuit 18. It is sent to F/F 31 provided in front of DDA in +1.

On the other hand, reading of the histogram data DM1 using the address signal β1 described above is executed in step 206. This step 206 corresponds to the address signal β in FIG.
This is done by applying (1) from the F/F 31 to the RAM, and storing the histogram data DMl at the address β in (2) the F/F 33. Then, in step 208, the gray value data DI of the histogram data DMl is added. This step 208 is executed in ADD of FIG. After that, the histogram data D'-aD N I added in step 208
+D r is written in the RAM at address β0 in step 210.

Refer to FIG. 10 to explain the accumulation of histogram data in more detail.

FIG. 10 shows the concept of a histogram memory using the RAM 34 in FIG.
-RAM, and these On-1 correspond to the rotation angles θ to θ of the recurrence formula calculation, respectively.
1 I am responding. And each RAM area is +512~0~-
It has 512 addresses β(-ρ), each address has 1
It is possible to store 6-bit histogram data. Therefore, in the DDA, 1+1 α, β corresponding to the rotation angle θ is calculated from the address signals α, β by the equation (7) mentioned above.
1+1 1
When the increment is 1, the address signal β becomes R
The histogram data D at the address β1 is read out. Then, the dark M of the point to be processed
(1) After being added to the light value data D1, the histogram data (D+D) is returned to address M(1) of RAM +.
1 Written to space β1.

As described above, the calculation of the recurrence formula shown in equation (7) is performed by the pipeline method by receiving and passing the address signals α and β1 one after another. And this α1. β to α
, β during the operation to I fil
Grayscale value data D of 141 togerum data D
Since accumulation (accumulation) is performed by l, the time required for the entire processing of one cycle can be shortened.

This one-cycle processing is performed in parallel by the DDA calculation circuits 18 to 18 forming the DDA calculation unit 18 when the circuits are open. In other words, when the data "■, empty, ■, empty, empty, ■, ■, empty" is input manually from the FIFO 17' as shown in FIG.
In the second cycle, it becomes as shown in (C) of the same figure, and in the third cycle, it becomes as shown in (d) of the same figure.The same process is performed thereafter, and in the 8th cycle, it becomes as shown in (D) of the same figure. e)
become that way. Here, (α 2 β
) to (α, β) are address signals p4° p4 corresponding to the coordinate values (xy) 1 4 pio p1 to (xy) of the processing target point p-p, respectively, and D1□ to DI4 are the processing target points P1-P4
This is the gradation value data for each. Also, the same figure (
θ to θ in b) to (e) is On-1 Rotation angle from the processing target point (θ. is the processing target point itself)
and correspond to RAMo to RAM in FIG. 2, respectively.

Next, neighborhood filtering after Hough transformation is completed will be described.

Now, when the intersection of the Rough curves is expressed on the ρ-θ plane, it is assumed that it becomes as shown in FIG. 12(a). Note that (a) in the same figure shows a histogram H in which the intersection points appearing in each unit of the ρ-θ plane are weighted by the gray value data (change rate of brightness) of the pixels (processing target points) in the original image. , are expressed using contour lines to make the explanation easier to understand, and do not necessarily match the histogram according to the present invention.

Here, the histogram is high at point p1 in FIG. It is assumed that the histogram is also high at p3. If we pay attention to the vicinity of point p1, we can see that there are also high histogram portions there, such as points p4 and p5. However, what is particularly important in image processing is to find points p1 to p3 that are far apart from each other; for example, point p1 is on the shoulder line of the road, point p2 is on the center line, and point p3 is on the shoulder of the curved road ahead. It corresponds to the line. On the other hand, the point p4.near the maximum histogram point pl. P5 and the like often correspond to partial bends in road shoulder lines, and in terms of image processing, they mainly correspond to noise components.

Therefore, the influence of such noise components is reduced by, for example, 8-neighborhood filtering. In other words, Fig. 12 (
An 8-neighborhood filter as shown in b) is prepared, and the histograms of the intersection points of the Hough curves for the area F-F9 are compared. Then, when F > F - F and F - F hold for the central area F5, the data of this area F5 is set as the data to be detected. Specifically, for example, when one unit (element area) pixel on the ρ-θ plane is assigned to F1 to F9, the histogram numbers of intersection points are F −6, F2-8, F3-4, ■F4-2, F5-14, F6■10, F7-7
, F8-9, F, -8, F
>F −F and F −F hold, so F5
Let the intersection point be the data to be detected. On the other hand, F -8, F2-4, F3-3, F4-14
, F5-10, F6-7, F7-9, F8-8,
When F9-2 is reached, since F5<F4, the area of F5 is not treated as data to be detected.

By performing the above filtering process, the 12th
In figure (a), point p4. The high points of the second and third histograms, p2., are unaffected by the presence of p5. p
3 can be detected.

That is, if the above filtering were not performed, the next high histogram point after the first high histogram point p1 would be point p4. p5, and the point p2. which is desired to be obtained as the second and third high histogram points. p3 becomes the fourth and fifth high histogram points, making subsequent signal processing extremely difficult.

Next, the sorting process shown as step 118 in FIG. 3 will be explained in detail with reference to FIG. 13.

FIG. 13 is a flow chart thereof. First, for sorting processing, input memory M1 and comparison memory MM
l～M Hn (n-1r 2 r 3 + 4・
) and initialize it (step 152)
.

Next, step 15 is to input data into the input memory Ml.
4) When this human power data is available (step 156)
Only then, the comparisons of steps 158, 162 and 166 are executed. Then, if the input memory M1 is larger,
Replace the contents with the corresponding comparison memo') M M (step 160°164.168). Then, the comparison memories MMl to MMn will eventually hold n pieces of manual data in descending order.

A concrete example of this is shown in FIG. 14.

First, four memories MMl'' MM4 are used as comparison memories.
Prepare. Then, the input data is “5, 7, 2
．． ．． 8, 4.9. The contents of comparison memories MMt"9", MM2-8", MM3-8, and MM4-7 are stored in . Note that this sorting process may be executed by software. By executing the series of processes as described above, all steps of signal processing by the image processing apparatus according to the present invention are completed. Then, an approximate straight line of the curve connecting the processing target points of the original image is found using the above values of ρ and θ.

The present invention is not limited to the above embodiments, and various modifications are possible.

For example, the window setting shown in step 124 in FIG. 5 may be performed for a plurality of windows, and furthermore, when the windows overlap, one of the windows may be processed preferentially.

Also, the window settings themselves are not necessarily necessary,
Hough transformation may be performed on the entire original image. However, in this case, the peripheral area of the original image indicated by the dotted line Q in Fig. 4(a) contains many noise components, so this area must be removed in advance when performing accurate image processing. It won't happen.

When obtaining a histogram of the intersection points of the Hough curves, only the concentration of the intersection points can be grasped as a histogram without superimposing shading value data regarding the rate of change in brightness (the value of the differentiated edged data). Furthermore, instead of converting the data into edge data through differentiation, the data corresponding to the luminance may be digitally processed as gradation value data, and then the histogram of the intersection of the Hough curves may be obtained.

It is not necessary to calculate the rotational motion recurrence formula for the entire circumference (-circumference) of the approximate circle, but for 1/2 circumference, 1/2 circumference, etc.
It may be 4 rounds or 1/8 round. For example 1/4
If the calculation is performed by arranging DDAs in series that perform circumference operations for 0≦θ×π/2 and π≦θ×3π/2, then other circumferences (π/2≦θ×π, 3π/2≦θku2
π) can be immediately determined from these. Further, it is not necessarily necessary that the rotation angles are always the same, and it is not impossible to make them partially different.

〔Effect of the invention〕

As explained above in detail, according to the present invention, D
The arithmetic unit can be easily formed by simply connecting DA arithmetic circuits in series, and it is not necessary to refer to a memory table or the like during arithmetic operations, so the configuration becomes extremely simple and compact. In addition, by configuring the calculation circuit to execute the rotational motion recurrence formula using the Vibrine method, it is possible not only to speed up calculations, but also to improve the initial value for calculating the rotational motion recurrence formula. This has the effect of eliminating the need for calculations.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the overall configuration of an image processing device according to an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of the main parts of FIG. 1, and FIG. 3 shows the overall operation. 4 is a diagram illustrating edge detection of an example of an original image and manually generated pixel signals, FIG. 5 is a flowchart illustrating preprocessing of edge data, and FIG. 6 is a diagram illustrating a lookup table. Fig. 7 is a diagram explaining the Ho u gh transformation in the embodiment of the present invention, Fig. 8 is a flowchart showing the calculation of the recurrence formula of rotational motion in the embodiment, and Fig. 9 is a diagram explaining the processing of one cycle. FIG. 10 is a conceptual diagram of the histogram memory, FIG. 11 is a diagram explaining the vibe line processing in the embodiment, FIG. 12 is a diagram explaining the 8-neighborhood filtering process, and FIG. 13 is a diagram explaining the sorting process. FIG. 14 is a diagram specifically explaining the sorting process, FIG. 15 is a diagram explaining road recognition, and FIGS. 16 to 18 are diagrams explaining the conventional Hou
It is a figure explaining gh conversion. 1... Camera image, 2... Horizon, 3... Road,
4... Road shoulder line, 5... Center line, 11.
. . . Camera, 12 . Pre-processing section, 17'-FIFOll 8-D D A calculation section,
18° ~ 18...DDA calculation circuit, 19... Neighborhood filter, 20... Sorting section, 21.23.
...VME bus, 22...CPU, 31, 32.33
...Flip-flop (F/F), 34-RAMo
~RAM (histogram memory), 35...A
DD -ADD (adder), 36...Buffer n-1 a, 37-D DA -D DA (DD
A K W times On-1 road). Patent Applicant: Honda Motor Co., Ltd. Representative Patent Attorney Yoshiki Hase Input Data
Input data (a) LUT L
(b) Lookup table with LUT (LU
Effect of T) Fig. 6 Fig. 8 Processing of one cycle Fig. 9 Concept of Hinutogram memory Fig. 10 Fig. 8 Neighborhood filtering Fig. 12 Fig. 15 Fig. 16 Procedure amendment document July 25, 1988 Nipponta Giken Columns and drawings of SanR Minami Kogyo Co., Ltd. 6 Contents of the amendment (1) "In this way...in this way" in lines 6 to 8 on page 10 of the specification is corrected to "in the former case." (2) On page 10, line 13, the ``documents of'' are corrected to ``the latter documents of.'' (3) rsinθ, cosθ on page 10, line 18
'' should be corrected as ``x-51nθ, y φcosθJ.'' p (4) Insert the following sentence next to ``Storage.'' in the first line of page 11. "Also, it is not suitable for LSI integration. A method of converting ROM to RAM may be considered, but this has the disadvantage of being placed next to the east." (5) Page 23, lines 3 and 4. “Kotomo・
··Recognize. " is corrected to "This is derived from the theorem of trigonometric functions." (6) In the 11th line of page 23 of the same figure, "point P(x. y)J is "p

The same figure (b) corresponding to θ-0@ of p p
The point Q (α.. β.)” is corrected. (7) DDA of [things] on page 24, line 9
Correct it with J. (8) “Toshito” on page 24, lines 11 to 17.
"..." should be corrected as shown below. [When closed, α −α −2βI country 1 β−2α+β ...(5) Ill fil f.”
...” to “Angle θ−0″ (i.e. ρ−X
from the position Q(α,β0)O corresponding to the angle).” (10) “Memory” on page 26, lines 5 and 12
Correct it to ``Remember as address.'' (11) Correct rRAM34J on page 26, line 15 as "RAM34J with address -β1." (12) Correct r+512J on page 32, line 6 to r+511.
Correct it to 4. (13) Correct the "input arrow" on page 37, line 7 to page 38, line 4 as shown below. ``Prepare a plurality of input memories (transfer memories) M1□ to M+4 and comparison memories (result memories) MML to MM4, each of which is assumed to be 4 to simplify the explanation, and initialize them (step 152). ).Next, input memory M1
1, step 154), and if this input data is found to be present in step 156, step 15
Execute steps 8 to 184, and if it is found to be none, proceed to step 19.
0 to 199 are executed. Here, steps 190, 196, and 199 are processed by steps 158, 16, respectively.
0, the processing of steps 192 and 198 are the same as the processing of steps 162-168, respectively, and the processing of step 194 is the same as the processing of steps 170-176. In steps 158, 162, 170, and 178, the contents of the corresponding input memory M1 and comparison memory MM are compared in size, and when M1≦MM, the contents of the input memory M1 are transferred to the next step (steps 164, 172, and 180). . On the other hand, when Ml>MM, the contents of comparison memory MM are transferred to the next input memory FA (step 16).
6゜174.182) At the same time, the contents of the manual memory M1 are put into the corresponding comparison memory MM (steps 1.68, 1
76.184). Then, finally the memories MMl~M
Four pieces of input data are held in M4 in descending order of size. A concrete example of this is shown in FIG. 14. First, four memories MIt to M+4 are used as input memories.
Four memories MM1 to MM4 are prepared as comparison memories. Then, the input data is 10 of "5.7, 2, 8, 4.9.3, 1.6°8" as shown in Figure 14 (cm).
Suppose that there are Then, by the operation after the initialization as shown in FIG. 3(b), the data stored in the comparison memories MM1 to MM4 are shown as arrows in FIG. 39(C). In the 13th line, "superimpose ... intersection point" is corrected to "by superimposing 21iff data instead of, the intersection point". (15) Insert the following sentence after "Not." on page 40, line 9. [Also, due to the properties of a straight line, 1/2 circumference (0” to 180°) may be subjected to Hough conversion.” ” he corrected. (17) "Arithmetic circuit" on page 42, line 7 is corrected to "arithmetic circuit and histogram memory." (18) Figures 7, 8, 10, 13 of the drawings,
Figures 14 and 16 are corrected as shown in the attached sheet. Figure 8 Concept of histogram memory Figure 10 Explanation of sorting Figure 14 (2) Figure 16

Claims (9)

[Claims] 1. In an image processing device that separates and extracts feature points of an original image from a plurality of processing target points on the original image captured by an imaging means, the DDA calculation means is configured by connecting a plurality of DDA calculation elements of the same configuration in series. and storage means for storing the calculation results of the DDA calculation means in correspondence with the DDA calculation elements, and means for extracting feature points of the original image based on the stored contents of the storage means. image processing device. 2. In an image processing device that derives an approximate straight line of a curve connecting multiple processing target points on an original image captured by an imaging means, the coordinates of a point on the circumference of an approximate circle drawn in an α-β orthogonal coordinate system are calculated as ( α_i, β_i), and the rotation angle to the coordinates (α_i_+_1, β_i_+_1) of the next point on the circumference is ε (where i is a positive integer), then α_i_+_1=f_α(α_i, β_i, ε )β_i
A DDA calculation means that sequentially calculates a rotational motion recurrence formula of ___+_1=f_β(α_i, β_i, ε) for each predetermined rotation angle in a pipeline method, and each result sequentially calculated by this DDA calculation means. Among them, a storage means for storing at least the value of β_i; and an approximate straight line deriving means for deriving the approximate straight line from the intersection of the Hough curves for each of the plurality of processing target points obtained based on the storage contents of the storage means. An image processing device comprising: 3. 3. The image processing apparatus according to claim 2, wherein the DDA calculation means is configured by connecting in series a plurality of calculation circuits that calculate the rotational motion recurrence formula for each rotation angle. 4. The rotational motion recurrence formula sets the rotational angle to ε=2^−
When ^m, α_i_+_1=α_i(1-2^-^2^m^-^1
)-2^-^mβ_iβ_i_+_1=2^-^mα_
The image processing apparatus according to claim 2, wherein i+β_i(1-2^-^2^m^-^1). 5. 3. The image processing apparatus according to claim 2, wherein the storage means stores at least the value of β_i in correspondence with the rotation angle. 6. 3. The image processing apparatus according to claim 2, wherein the storage means stores gradation value data corresponding to the brightness of the processing target point or the rate of change thereof in the original image. 7. 6. The image processing apparatus according to claim 5, wherein the storage means stores the grayscale value data of the processing target point in the original image in correspondence with the rotation angle. 8. 3. The image processing apparatus according to claim 2, wherein the approximate straight line deriving means includes a neighborhood filter that performs neighborhood filtering processing on data regarding the intersection of the Hough curves. 9. 7. The image processing apparatus according to claim 6, wherein the approximate straight line deriving means includes a neighborhood filter that superimposes the gradation value data on data regarding the intersection of the Hough curve and performs neighborhood filtering processing.