JP3724659B2 - Image processing device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing apparatus that processes a two-dimensional image signal (image frame signal) output from an imaging means attached to a robot hand, for example, and obtains an image signal for performing inspection, object recognition, device operation control, and the like. About.
[0002]
[Prior art]
Japanese Patent Application Laid-Open No. Sho 61-293657 is caused by illumination conditions that cannot be avoided with a single image frame signal under a single illumination condition (exactly, the background in the imaging region or the reflection state of the imaging object). In order to reduce unnecessary signals, a plurality of light sources that are arranged in different spatial positions and illuminate an imaging target, a control unit that individually controls lighting of each light source, and a different lighting condition of each light source An image pickup means for sequentially outputting a plurality of image frame signals obtained by sequentially picking up an image pickup object in step 1, an image signal processing means for binarizing each image frame signal, and a plurality of images output from the image signal processing means An image processing apparatus (image inspection apparatus) is provided that includes a determination unit that determines the quality of an imaging target based on the binarized image signal.
[0003]
The processing method of a plurality of binarized image signals specifically disclosed in the above publication is as follows. That is, the area, circumference, and vertical or horizontal length of the white area (or black area) in each binary image frame signal is calculated and compared with a predetermined threshold value.
[0004]
[Problems to be solved by the invention]
The discriminant method of the above publication is a method for determining whether or not a plurality of image frame signals are captured by changing the illumination direction as described above, and finally making a final determination based on the determination results. Judgment is inevitably determined to be finally good when all the determinations of each image frame signal are good, and after all, it is merely inspected a plurality of times by changing the illumination light irradiation direction. Absent.
[0005]
However, simply by changing the illumination light irradiation direction and inspecting a plurality of times, the image to be imaged (inspection image) or background image due to unnecessary reflected light or unnecessary shadows due to the illumination light irradiation direction. It is difficult to remove unnecessary signals that are inevitably mixed therein, and image determination must be performed on the premise of the presence of these unnecessary signals.
Of course, in order to avoid unnecessary reflection (particularly reflection of the light source on a smooth surface) and generation of shadows, it is possible to devise the use of a surface light source, but in the inspection of the imaging target, light is captured from a predetermined direction such as oblique light. There are cases where it is desired to enhance the outline by irradiating the object, or there is a case where a spot light source must be employed in the layout at the production site. In such a case, generation of the above-described unnecessary reflection or shadow occurs.
[0006]
Further, the multiple inspection method disclosed in the above publication has the disadvantages that the circuit scale of an image signal processing unit such as an image memory becomes large and the processing time becomes long.
Specifically, for example, when illuminating the periphery of the imaging target from a certain oblique angle, if oil adheres to the surface of the imaging target or the surface of the support base on which the imaging target is placed, Illumination light to be scattered may be totally reflected and enter the TV camera, and a wrinkle region that should originally be a black region may become a white region. Such false signal removal processing is cumbersome and requires complex image processing.
[0007]
Further, depending on the illumination direction as well as the oil surface, unnecessary reflected light may be generated on the imaging target and the mounting table itself on which the imaging target is mounted, resulting in the same problem as described above.
Furthermore, the above-described generation of unnecessary reflections and shadows is a problem not only in image inspection but also in image recognition by capturing an object.
The present invention has been made in view of the above problems, and provides an image processing apparatus capable of accurately removing unwanted signals due to illumination conditions without using complicated image processing and a large-scale apparatus. This is the first purpose.
[0015]
[Means for Solving the Problems]
In the first configuration of the present invention,First and second light sources are disposed on both sides of the imaging unit, and third and fourth light sources are disposed on both sides of the imaging unit at positions different from those. The first and second image frame signals obtained by turning on the first and second light sources during the first imaging period and turning on the third and fourth light sources during the second imaging period are obtained for each pixel. The data is converted into a logical product for each pixel. If it does in this way, there will be the following effects.
[0016]
Although it is extremely advantageous from a practical point of view that the camera (imaging means) and the light source are arranged in the robot hand in that the hand does not enter the field of view of the camera or the illumination field of the light source, the viewpoint of securing the freedom of movement of the hand Therefore, it is extremely difficult to install a uniform illumination light source or a large planar light source on the hand, and it is extremely difficult to use a hand with a large illumination light source. A point light source is most preferred.
[0017]
However, when the radiated light from such a point light source (non-uniform illumination light source) is reflected by a smooth surface such as a ground surface or an oil surface thereon, even if the same plane is imaged, a certain area (hereinafter referred to as the region) In this case, the brightness of the reflected light becomes extremely high (the part that is originally black as the binarized signal becomes white). This problem peculiar when light source (non-uniform illumination light source) light is used as illumination light is obtained by binarizing two image frame signals, which are signals of different imaging periods, for each pixel and then taking a logical product of them. This is solved as described above.
[0018]
However, if a point light source is placed on one side of the camera, the amount of reflected light incident on the camera from the work surface opposite to the point light source across the camera is reflected from the amount of reflected light incident on the camera from the work surface on the camera side. It will decrease more significantly. Further, if there is unevenness or a step in the work, a shadow is generated, and the pixel signal in this shadow portion becomes black due to the logical product, and a normal work image cannot be obtained.
[0019]
In order to reduce shadows and reduce variations in the amount of reflected light to the camera due to the positional relationship with the point light source, multiple point light sources may be provided around the camera and lighted at the same time for imaging. . However, if this is done, the number of shining (high bright spot) regions in the first and second image frame signals increases, so the possibility that the shining regions of both image frame signals overlap will increase. . If this overlap occurs, the above-mentioned logical product process cannot remove the margin area.
[0020]
That is, if the number of simultaneously lit light sources provided in the robot hand is small, useless shadows and variations in brightness in each part of the work (as viewed from the camera) occur. If the number of these simultaneously lit light sources increases, The key (high bright spot) area overlaps and cannot be removed by AND processing.
This configuration is made to remedy this problem, and the first and second light sources that are turned on simultaneously are arranged on both sides of the image pickup means, and they are located at positions different from those (preferably at positions perpendicular to them). The third and fourth light sources that are turned on simultaneously are arranged on both sides of the image pickup means. In this way, it is possible to effectively reduce the shadows in both image frame signals, and further reduce the overlap between the areas of the two image frame signals (high bright spot regions). Can be obtained.
[0021]
  Further, in this configuration, since the scale can be removed by binary information processing called logical product operation, the storage and processing of each pixel signal is simplified.
  First of the present invention2With this configuration, a light source is provided on the side of the imaging means at the tip of the robot hand, and an image frame signal obtained by turning on the light source is binarized for each pixel to obtain a binarized image frame signal. In particular, in this configuration, threshold levels are individually stored in advance for each pixel, and the image frame signal is binarized for each pixel using the threshold level. If it does in this way, there will be the following effects.
[0022]
Although it is extremely advantageous from a practical point of view that the camera (imaging means) and the light source are arranged in the robot hand in that the hand does not enter the field of view of the camera or the illumination field of the light source, the viewpoint of securing the freedom of movement of the hand Since it is quite a burden to install a camera alone, it is extremely difficult to use a uniform illumination light source or a large planar light source on the hand. A point light source is most preferred.
[0023]
However, the reflected light emitted from such a point light source (non-uniform illumination light source), reflected from the workpiece or background and incident on the camera is the largest when reflected from the midpoint between the camera and the light source. As the distance from the optical axis increases, the amount of reflected light that enters the camera varies.
This configuration is made in order to solve this problem in the case of a light source provided in a robot hand. Since the binarization process is performed for each pixel according to a threshold level stored in advance, the above-described incident light amount is the pixel. The problem of variation every time is solved.
[0024]
  First of the present invention3In this configuration, the first light source disposed on one side of the imaging unit and the second light source disposed on the other side of the imaging unit (a side different from the one side) are sequentially turned on to perform the first. Then, the second image frame signal is sequentially taken out, the two image frame signals are binarized for each pixel, and a logical product operation is performed for each pixel to remove the above-described amount of light. In particular, in this configuration, since the signals of the same pixel of both image frame signals are individually binarized at different threshold levels, the following operational effects are obtained.
[0025]
Although it is extremely advantageous from a practical point of view that the camera (imaging means) and the light source are arranged in the robot hand in that the hand does not enter the field of view of the camera or the illumination field of the light source, the viewpoint of securing the freedom of movement of the hand Therefore, it is very difficult to install a large surface light source on the hand, and it is extremely difficult to use a large surface light source on the hand. A point light source such as a small spotlight is the most suitable light source for the robot hand. Is preferred.
[0026]
However, when the radiated light from such a point light source is reflected by a smooth surface such as a ground surface or an oil surface thereon, even if the same plane is imaged, a certain region (hereinafter, also referred to as a telemetry region) Then, the lightness becomes extremely high by the incidence of the reflected light (the part that is originally black as the binarized signal becomes white). This problem peculiar when using point light source (non-uniform light source) light as illumination light is that two image frame signals that are signals of different imaging periods are binarized for each pixel, and then the logical product of them is calculated. This is solved as described above.
[0027]
However, since the positions of the first light source and the second light source are different, the amount of reflected light that is reflected from the light source at a predetermined point in the imaging region and is incident on the camera (imaging means) is greatly different and has different binary levels. It turned out that it might be converted.
This configuration has been made to solve this problem. Different threshold levels are stored in advance for the same pixel of both image frame signals, and both image frame signals are stored according to these threshold levels. The above problem is solved by binarizing separately.
[0028]
  First of the present invention4In this configuration, the first and second image frame signals are obtained by sequentially lighting the first light source disposed on one side of the imaging unit and the second light source disposed on the other side of the imaging unit. The two image frame signals are sequentially extracted, binarized for each pixel, and subjected to a logical product operation for each pixel to remove the fog as described above. In particular, in this configuration, a state quantity related to the average signal level of the entire image frame signal is extracted, and the state quantity is set in a direction in which the number of white pixels in the binarized image frame signal approaches a predetermined reference number. Based on this, the threshold level for each pixel is shifted by the same amount. If it does in this way, there will be the following effects.
[0029]
Although it is extremely advantageous from a practical point of view that the camera (imaging means) and the light source are arranged in the robot hand in that the hand does not enter the field of view of the camera or the illumination field of the light source, the viewpoint of securing the freedom of movement of the hand Since it is quite a burden to install a camera alone, it is extremely difficult to use a uniform illumination light source or a large planar light source on the hand. A point light source is most preferred.
[0030]
However, when a workpiece is imaged with a set of a camera and a light source fixed to the robot hand as described above, the point light source is radiated from the point light source due to slight changes in the posture of the workpiece and the posture of the hand. The amount of reflected light that is reflected and incident on each pixel of the camera (imaging means) may change uniformly. For example, if the reflecting surface of the work is slightly inclined with respect to the optical axes of the parallel camera and the point light source, the amount of reflected light incident on the camera from the reflecting surface will vary slightly for each pixel. It changes uniformly.
[0031]
  This configuration is made to solve this problem, and extracts the amount of change in the average signal level of the entire image frame signal, and accordingly shifts the threshold level for each pixel by the same amount, This change in average signal level is canceled (compensated).
In this configuration,Since the threshold level is changed so that the number of white pixels in the binarized image frame signal becomes a predetermined reference value, the overall change in the amount of reflected light incident on the camera as described above can be easily fixed. In addition, even when a large amount of reflected light is incident from a part of the pixels as in the case of the above-mentioned balance, it is possible to prevent the problem that this changes the overall change in the amount of reflected light.
[0033]
DETAILED DESCRIPTION OF THE INVENTION
Example 1
(Configuration explanation)
An embodiment of the image processing apparatus of the present invention will be described with reference to FIGS. FIG. 1 is a side view showing a main part of the image processing apparatus, FIG. 2 is a layout diagram showing a planar arrangement of a light source device 10 used in the image processing apparatus, and FIG. 3 is a block circuit diagram of the image processing apparatus. FIG. 4 is an explanatory diagram for explaining a logical product process, which is a feature of the image processing apparatus.
[0034]
As shown in FIG. 3, the image processing apparatus performs an image processing on an image frame signal output from the TV camera (imaging means) 1, the light sources 21 to 24, and the TV camera 1 and outputs a processed image signal. A signal processing circuit (image signal processing means) 3, an image signal processor 4 that performs signal processing on the processed image signal to determine whether the imaging target is acceptable, and a timing controller (lighting control means) 5 are provided.
[0035]
A TV camera 1 having an optical axis that hangs vertically images a press punched product (imaging target) 7 having a predetermined three-dimensional shape on a pallet 6, and light sources 21 to 24 on the side, front, rear, left and right of the TV camera 1. Are arranged individually. More specifically, the TV camera 1 is formed integrally with the light sources 21 to 24 and can move relative to the press punched product 7 on the pallet 6 in the front-rear and left-right (horizontal direction) directions. Further, it is assumed that the oil film 8 is attached to the upper surface of the pallet 6 and the oil film 80 is also attached to the upper surface of the press punched product 7.
[0036]
The image signal processing circuit 3 binarizes the image frame signal output from the TV camera 1 for each pixel and outputs the binarized image frame signal, and the binarized image frame signal output from the comparator 31. Frame memory 32 for storing only one frame, and an AND circuit that performs a logical product operation for each pixel of the read binary image frame signal read from the frame memory 32 and the binary image frame signal output from the comparator 31 33.
[0037]
The image signal processor 4 calculates a predetermined inspection amount from the input signal based on the built-in image processing program and determination program, but the determination routine itself is not the gist of the present embodiment, so that further detailed description will be given. Is omitted.
The timing controller 5 is a circuit that controls the timing of the TV camera 1, the light sources 21 to 24, and the frame memory 32, but a specific circuit description thereof is omitted.
[0038]
The light sources 21 to 24 are arranged at a right angle to each other and 90 degrees apart from each other with the optical axis of the TV camera 1 as a center, and are controlled by the timing controller 5 to be a light source in a first imaging period for obtaining a first image frame signal. 21 and 23 are turned on, and the light sources 22 and 24 are turned on in the second imaging period for obtaining the second image frame signal. In the present embodiment, the light sources 21 and 23 are symmetrically arranged with the TV camera 1 in between, and therefore, the shadow of the press punched product 7 existing immediately below the optical axis of the TV camera 1 does not easily occur due to the lighting thereof. Similarly, since the light sources 22 and 24 are symmetrically arranged with the TV camera 1 in between, the shadow of the press punched product 7 existing immediately below the optical axis of the TV camera 1 is less likely to occur due to their lighting.
[0039]
(Description of operation)
When the TV camera 1 is stopped immediately above the pallet 6, a start signal is output from an external controller (not shown) to the timing controller 5, and the timing controller 5 outputs the left and right light sources 21 of the TV camera 1 based on the start signal. , 23 is lit, and the TV camera 1 is instructed to image the upper surface of the pallet 6 and the press punched product 7 placed on the pallet 6, and the TV camera 1 serially outputs the first image frame signal A for one frame. Output. This signal output is executed in synchronism with the vertical and horizontal synchronizing signals from the timing controller 4 to the TV camera 1, and the first image frame signal is binarized by a comparator 31 having a predetermined threshold value, and the frame memory 32. Stored in Note that during this imaging period, the timing controller 4 outputs a lighting signal W, and the lighting is performed.
[0040]
Thereafter, when the storage of the binary image frame signal in the frame memory 32 is finished, the timing controller 4 instructs the TV camera 1 to take an image again while turning on the light sources 22 and 24 before and after the TV camera 1. The TV camera 1 serially outputs the second image frame signal B for one frame as described above, and the second image frame signal B is also binarized by the comparator 31.
[0041]
The timing controller 4 outputs the read signal R and the vertical and horizontal synchronization signals to the frame memory 32 when the image frame signal is output, whereby the frame memory 32 outputs the first binarized image frame signal A from the comparator 31. Read out at the same pixel timing as the output binarized image frame signal B.
The two binarized image frame signals A and B output in synchronization with each other in this manner are ANDed for each pixel by the AND circuit 33, and the output, that is, the logical product of the binarized image frame signals. The signal is output to the image signal processor 4 for further processing.
[0042]
Next, the effect of the above-mentioned AND operation will be described with reference to FIG.
4A shows the image A represented by the first binarized image frame signal, and 71 and 72 are images of the light sources 21 and 23 concentratedly reflected by the oil film 8 attached to the upper surface of the pallet 6. 4B shows an image B represented by the second binarized image frame signal, and 73 and 74 are images of the light sources 22 and 24 concentratedly reflected by the oil film 8 attached to the upper surface of the pallet 6.
[0043]
Since the light sources 21 to 24 are arranged apart from each other, the images 71 to 74 are formed at different portions on the oil film 8. Therefore, by performing an AND operation on the image a and the image b for each same pixel, the concentrated reflection due to the oil film 8 can be removed very easily as shown in FIG.
(Modification)
FIG. 5 shows different arrangement examples of the light sources 21 to 24. (A) arrange | positions each light source 21-24 at each vertex of the rectangle which makes TV camera 1 a gravity center, (b) cross | intersects each light source 21-24 at right angle with the optical axis M of TV camera 1. FIG. It is arranged on the horizontal line. In addition, if the combination of the light sources that are sequentially turned on is line-symmetrical with respect to the optical axis M (that is, provided on both sides of the TV camera 1), the number, shape, and arrangement pattern of the light sources are also free. It is. For example, each of the light sources 21 to 24 may be a linear light source such as a fluorescent lamp or a ring light source, in addition to a light source such as a spotlight, or a ring light source arranged coaxially.
(Example 2)
Another embodiment will be described with reference to FIGS.
[0044]
In this embodiment, the light sources 21 to 24 are always turned on to illuminate the entire pallet 6. The TV camera 1 is moved independently of the light source in the horizontal direction, forward and backward, left and right or both by an actuator (not shown) such as a robot hand.
For example, the TV camera 1 is moved forward to obtain the image a, and the TV camera 1 is moved backward to obtain the image b. In this way, since the reflection positions of the light sources 21 to 24 on the surface of the oil film 8 are moved, the reflection images of the light sources 21 to 24 can be removed from the logical product image c.
[0045]
However, in this embodiment, the coordinate position of the image frame signal is moved in the same direction or in the opposite direction by the movement of the TV camera 1, so that the entire screen is shifted in the opposite direction to form both image frame signals. The pixel coordinate positions of the images of the two punched press products (imaging targets) in the pair of images are matched. That is, control is performed so that the same pixel signals of both images a and b are input to the AND circuit 33 at the same timing.
[0046]
In this embodiment, this image shift is performed by adjusting the read timing of the stored image frame signal from the frame memory 32.
That is, if the vertical direction, the horizontal direction, and the direction of the imaging surface of the TV camera 1 at the two imaging positions coincide with each other, the TV image can be easily obtained by adjusting the read timing of the stored image frame signal from the frame memory 32. Pixel coordinate positions on the virtual image space to be imaged as the camera 1 moves can be matched.
[0047]
Further, according to this embodiment, the logical product of both the image frame signals a and b is obtained. In this way, the following effects can also be obtained. That is, when binarization is performed so that the background is black (0) and the imaging target is white (1), if the imaging position of the TV camera 1 is different, the shape of the upper surface of the imaging target when the imaging target is a three-dimensional image. Are identical in both image frame signals, but because the appearance of the side surface (wall surface) is different, the shape of the imaging target is different in both image frame signals. However, in this embodiment, since the logical product of both image frame signals is taken, the side surface of the imaging target that becomes a white area at different positions becomes black (0) by the logical product operation and can be removed, so-called parallax. It is possible to remove the shape difference of the imaging target associated with.
[0048]
Also in the present embodiment, the lighting light source at the time of imaging at each imaging position can be changed. That is, when the TV camera 1 captures an image at the first imaging position, the first group of lamps is turned on, and when the TV camera 2 captures an image at the second imaging position, the second group of lamps is lit. Of course, the first group of lamps and the second group of lamps have different lighting positions. In this way, if the logical product processing of both the image frame signals a and b is executed for each pixel, in some cases, it is possible to remove an unnecessary reflection image input to the TV camera even better.
(Example 3)
Another embodiment will be described with reference to FIG.
[0049]
In this embodiment, the AND circuit 33 in FIG. 3 is changed to a comparator 334 and analog gate circuits 335 and 336, and the position of the comparator 31 is changed. The frame memory 32 is an 8-bit (256 gray scale) image memory, and an A / D converter (not shown) is provided at the input end, and a D / A converter (not shown) is provided at the output end. Is installed.
[0050]
The operation of this circuit will be described below.
After the first image frame signal (image) a having 8-bit light and dark information is written to the frame memory 32 by the same writing operation as in the first embodiment, the image frame signal (image) b which is an analog signal from the TV camera 1 is written. Are output, and the images a and b are compared by the comparator 334 in order for each pixel. When the pixel signal of the image a is large (when bright), the comparator 334 outputs a high level to turn on the analog gate 335 and cut off the analog gate 336. As a result, the pixel signal of the image b which is the darker pixel signal is selected.
[0051]
Similarly, when the pixel signal of the image a is small (when dark), the comparator 334 outputs a low level, shuts off the analog gate 335, and turns on the analog gate 336. As a result, the pixel signal of the image a which is the darker pixel signal is selected. The signals of both analog gates are binarized by the comparator 31 and output to the circuit 4.
In this way, the same effect as Example 1 can be produced.
Example 4
Another embodiment will be described with reference to FIGS.
[0052]
This embodiment is a modified embodiment of the second embodiment. Instead of moving the TV camera 1 with a robot hand or the like as described above, one TV camera 1a or 1b is individually arranged at two imaging positions in advance. It is to be established. In this way, the same effects as in the second embodiment can be obtained.
The operation of this image processing apparatus will be described with reference to FIG. 9. Analog image frame signals output from the TV cameras 1a and 1b are individually binarized by binarization circuits 31a and 31b, respectively, and binarized images are obtained. The frame signal a is ANDed with the binarized image frame signal b delayed for a predetermined time by the delay circuit 32 a and the AND circuit 4.
[0053]
Since the circuit operation and the operational effect itself are the same as those in the second embodiment, further explanation is omitted.
In this embodiment, the delay circuit 32a is a variable delay circuit, and the delay time can be appropriately changed in accordance with a change in the imaging position of the TV camera 1 and the like.
(Example 5)
Another embodiment will be described with reference to FIGS.
[0054]
This embodiment is a modified embodiment of the fourth embodiment. Instead of providing two TV cameras 1, an imaging surface of one TV camera 1 using a double periscope type mirror device (or prism device) 9 is used. The signal light similar to that captured at two imaging positions is incident on the right and left of the. In the mirror device 9, reference numeral 90 denotes a rectangular tube barrel, and 91 to 94 denote mirrors.
[0055]
FIG. 11 shows the imaging surface 100 of the TV camera 1. The signal light incident from the mirror 91 is incident on the left half 100L of the imaging surface 100, and the signal light incident from the mirror 93 is incident on the right half 100R of the imaging surface 100.
The operation of this image processing apparatus will be described with reference to FIG.
The image frame signal output from the TV camera 1 is binarized by the binarization circuit 31, and then the binarized image frame signal b delayed for a predetermined time by the delay circuit 32 and the binarized image frame signal not delayed. The AND of these binarized image frame signals a and b is taken by the AND circuit 4.
[0056]
Since the circuit operation and the operational effect itself are the same as those in the second embodiment, further explanation is omitted.
(Example 6)
Another embodiment will be described with reference to FIGS. In the present embodiment, a TV camera (imaging means) 1, four surrounding light sources 21 to 24, a timing controller 5 that drives and controls the light sources 21 to 24, and an image frame signal output from the TV camera 1 are pixelated. An A / D converter 10 that performs A / D conversion every time, and a microcomputer (image signal processing means, threshold) that stores a digital image frame signal output from the A / D converter 10 and outputs a processed image signal. Value level storage means, overall brightness normalization means) 30. The timing controller 5 is controlled by the microcomputer 30, and the TV camera 1 outputs a horizontal synchronization signal and a vertical synchronization signal to the microcomputer 30.
[0057]
Hereinafter, the operation of the microcomputer 30 will be described with reference to the flowcharts of FIGS.
First, the initial setting operation of each part is performed by turning on the power (100), and it is checked whether an imaging command from the outside is input (102), and waits until it is input. The first and second light sources 21 and 22 arranged on the TV are turned on (104), and after a predetermined time (105) when the illuminance is stabilized from the lighting command in step 104, the TV camera is synchronized with the vertical synchronization signal. The first image frame signal S1, which is a digital image signal of 1 to 1 frame, is read and stored in the first frame memory (106), and the first and second light sources 21 and 22 are turned off (107).
[0058]
Next, the third and fourth light sources 23 and 24 arranged on both sides of the TV camera 1 are turned on (108), and after a predetermined time (109) when the illuminance is stabilized from the lighting command of step 108, the vertical direction is reached. In synchronization with the synchronization signal, the second image frame signal S2 which is a digital image signal of one frame is read from the TV camera 1 and stored in the second frame memory (110), and the third and fourth light sources 23 and 24 are turned off. (112). The light sources 23 and 24 are disposed at right angles to the light sources 21 and 22 with the camera optical axis as the center.
[0059]
Here, the microcomputer 30 has a built-in first threshold value memory, and the first threshold value memory has a first threshold value for binarizing the first image frame signal S1 for each pixel. Assume that the value signal VtS1 is stored.
Next, the pixel signal S1PCi at the i-th address of the first image frame signal S1 is read from the first frame memory, and the pixel signal VtS1i at the i-th address of the first threshold signal VtS1 is read from the first threshold memory in synchronism therewith. Read and compare the magnitudes of both (200). Note that i is initially 0.
[0060]
If S1PCi is larger than VtS1i, 1 is stored at address i of the first binary image memory (202), and the process proceeds to step 206. If S1PCi is not greater than VtS1i at step 200, 0 is stored at the address i of the first binary image memory (203), and the process proceeds to step 206.
In step 206, 1 is added to the number of addresses i, and then it is checked whether i is the last address of the imaging area (first binary image memory) (208). If not, the process returns to step 200. If i is the last address of the imaging area, i is reset to 0 (210), and the process proceeds to step 212.
[0061]
Next, the pixel signal S2PCi at the address i of the second image frame signal S2 is read from the second frame memory, and the pixel signal VtS2i at the address i of the second threshold signal VtS2 is read from the second threshold memory in synchronization therewith. Read and compare the magnitudes of both (212).
If S2PCi is larger than VtS2i, 1 is stored at address i of the second binary image memory (214), and the process proceeds to step 218. If S2PCi is not larger than VtS2i at step 212, 0 is stored at address i of the second binary image memory (215), and the process proceeds to step 218.
[0062]
In step 218, 1 is added to the number of addresses i, and then it is checked whether i is the last address of the imaging area (second binary image memory) (220). If not, the process returns to step 212. If i is the final address of the imaging area, i is reset to 0 (222), and the process proceeds to step 400.
Hereinafter, the first binarized image frame signal stored in the first binary image memory is referred to as S1 ′, the value of the i address is referred to as S1PCi ′, and the first binary image frame signal stored in the second binary image memory is referred to as S1PCi ′. The binary image frame signal is called S2 ′, and the value at the i address is called S2PCi ′.
[0063]
Step 400 is a logical product of the first binarized image frame signal S1PCi ′ at address i binarized with a suitable amount of reflected light by the above routines and the second binarized image frame signal S2PCi ′ at address i. In this step, if both S1PCi ′ and S2PCi ′ are 1 (white), 1 is written to the i address of the first binarized image memory (402), and the built-in accumulated white address 1 is added to the count value N of the white counter for counting the number, and the process proceeds to step 406. If both S1PCi ′ and S2PCi ′ are 1 (white), 0 is set to the address i of the first binarized image memory. Write (404), add 1 to i (406), check whether i is the last address iend (408), otherwise return to step 400, otherwise go to step 300.
[0064]
In step 300, it is checked whether or not the accumulated number of white addresses N is within a pre-stored preferred range (Nmin <N <Nmax). If N is equal to or less than the predetermined minimum value Nmin, the threshold value is totally set. If it is high (the amount of reflected light is too small as a whole), it is determined to be inappropriate, and the routine proceeds to step 302.
In steps 302 to 308, pixel signals VtS1i and VtS2i (i = 0 to iend) at all addresses of the first threshold signal VtS1 and the second threshold signal VtS2 stored in the memory for storing threshold values are respectively obtained. This is a routine for reducing by a small constant value ΔVt. First, ΔVt is subtracted from the two pixel signals VtS1 and iVtS2i at address i (302), 1 is added to i (304), and whether i reaches iend (306), if not reached, the process returns to step 302. If reached, i is reset to 0 (308), and the process returns to step 200.
[0065]
If N is greater than or equal to the predetermined maximum value Nmax in step 300, the threshold value is generally low (the amount of reflected light is generally too large), and it is determined that it is inappropriate, and the process proceeds to step 312.
Steps 312 to 318 have essentially the same routine structure as Steps 302 to 308 described above, and pixel signals VtS1i and VtS2i (i = 0) at all addresses of the first threshold value signal VtS1 and the second threshold value signal VtS2. .About.iend) is incremented by a small constant value .DELTA.Vt, and further explanation is omitted.
[0066]
Further, if N is within the above range in step 300, the routine is terminated.
As a result, the third binary image memory removes the scale using the light / dark image binarized with the optimum threshold value for each pixel as described above, so that an accurate work shape can be obtained.
[Brief description of the drawings]
FIG. 1 is a side view illustrating a main part of an image processing apparatus according to a first embodiment.
2 is a layout diagram of light sources used in the image processing apparatus of FIG. 1. FIG.
3 is a block diagram of the image processing apparatus of FIG. 1. FIG.
FIG. 4 is an explanatory diagram illustrating image processing according to the first exemplary embodiment.
FIG. 5 is a layout diagram showing a modification of the light source layout of FIG. 2;
FIG. 6 is a perspective view illustrating a main part of the image processing apparatus according to the second embodiment.
FIG. 7 is a logic circuit diagram showing a main part of a third embodiment.
FIG. 8 is a front view illustrating a main part of an image processing apparatus according to a fourth embodiment.
FIG. 9 is a block circuit diagram showing a main part of a fourth embodiment.
FIG. 10 is a partially broken front view showing a main part of an image processing apparatus according to a fifth embodiment.
FIG. 11 is a diagram illustrating an imaging surface of the TV camera 1 according to the fifth embodiment.
FIG. 12 is a block circuit diagram showing a main part of the fifth embodiment.
13 is a block circuit diagram showing Example 6. FIG.
14 is a flowchart showing the operation of the circuit of FIG.
15 is a flowchart showing the operation of the circuit of FIG.
16 is a flowchart showing the operation of the circuit of FIG.
17 is a flowchart showing the operation of the circuit of FIG.
[Explanation of symbols]
Reference numeral 1 denotes a TV camera (imaging means) 21 to 24, a light source, 3 an image signal processing circuit (image signal processing means), 5 a timing controller (control means), and 7 a press punched product (imaging target).

Claims (4)

An imaging unit that is fixed to the tip of the robot hand and sequentially outputs first and second image frame signals obtained by imaging the imaging target in the first and second imaging periods, respectively;
First and second light sources that illuminate the object to be imaged while being fixed to the tip of the robot hand while being close to both sides of the imaging means;
Third and fourth light sources that illuminate the imaging target by being fixed to the tip of the robot hand while being close to both sides of the imaging means at a site different from the first and second light sources,
The first and second light sources are turned on during the first imaging period, the third and fourth light sources are turned off, and the third and fourth light sources are turned on during the second imaging period. Lighting control means for turning off the first and second light sources;
Image signal processing means for binarizing the first and second image frame signals for each pixel and performing a logical product operation for each pixel;
An image processing apparatus comprising: An imaging means for sequentially outputting image frame signals obtained by sequentially imaging an imaging target fixed to the tip of the robot hand; and an imaging means fixed to the tip of the robot hand in proximity to the imaging means. In an image processing apparatus comprising: a light source that illuminates; and an image signal processing unit that binarizes the image frame signal for each pixel at a predetermined threshold level.
The image signal processing means individually sets the threshold level for each pixel according to the amount of reflected light that is emitted from the light source , reflected from a work or background and incident on the imaging means varies for each pixel. Threshold level storage means for storing, and binarization means for comparing each of the threshold levels read from the threshold level storage means with the image frame signal for each pixel. An image processing apparatus. An imaging unit that is fixed to the tip of the robot hand and sequentially outputs first and second image frame signals obtained by imaging the imaging target in the first and second imaging periods, respectively;
A first light source that is fixed to the tip of the robot hand and illuminates the imaging object, in proximity to the imaging means;
A second light source that illuminates the imaging target by being fixed to the tip of the robot hand while approaching the imaging means at a site different from the first light source;
Turn on the first light source and turn off the second light source during the first imaging period, turn on the second light source and turn off the first light source during the second imaging period Control means;
Image signal processing means for performing a logical product operation for each pixel after binarizing the first and second image frame signals for each pixel at a predetermined threshold level;
The image signal processing means has a reflected light quantity that is reflected at a predetermined point in the imaging region from the two light sources and incident on the imaging means due to the different positions of the first light source and the second light source. the image processing apparatus characterized by separately binarized at different threshold levels signals of the same pixel of the two images frame signal in accordance with the difference. An imaging unit that is fixed to the tip of the robot hand and sequentially outputs first and second image frame signals obtained by imaging the imaging target in the first and second imaging periods, respectively;
A first light source that is fixed to the tip of the robot hand and illuminates the imaging object, in proximity to the imaging means;
A second light source that illuminates the imaging target by being fixed to the tip of the robot hand while approaching the imaging means at a site different from the first light source;
Turn on the first light source and turn off the second light source during the first imaging period, turn on the second light source and turn off the first light source during the second imaging period Control means;
Image signal processing means for performing a logical product operation for each pixel after binarizing the first and second image frame signals for each pixel at a predetermined threshold level;
The image signal processing means extracts the number of white pixels in the binarized image frame signal, and the direction in which the number of white pixels in the binarized image frame signal approaches a predetermined reference number An image processing apparatus comprising overall brightness adjustment means for shifting the threshold level for each pixel by the same amount.

Images