JP6199690B2 - Image processing system and work robot - Google Patents

Description

  The present invention relates to an image processing system that performs communication on captured pixel data and a working robot that operates based on pixel data acquired by the image processing system.

  2. Description of the Related Art Conventionally, there are work robots, for example, electronic component mounting apparatuses, which include an imaging device for imaging a circuit board on which electronic components are mounted (for example, Patent Document 1). In the electronic component mounting apparatus, for example, pixel data captured by the imaging apparatus is transmitted to the control apparatus via a communication cable. The control device performs image processing on the input pixel data, detects an error in the holding position of the circuit board, etc., adjusts the position of the mounting head that holds the electronic component based on the detection result, and performs the mounting operation. In communication between the imaging device and the control device, for example, data is transmitted according to a communication standard based on the CameraLink standard using LVDS (Low Voltage Differential Signaling) technology. The CameraLink standard here is one of the communication standards that define the data transmission system of industrial digital cameras, and is a communication standard that standardizes connectors and pin assignments and transmits pixel data and the like by serial communication.

  In addition, even if the above-described imaging apparatus is an apparatus of the same standard, an error occurs in the mounting position of a lens or an imaging element (for example, CMOS) due to a limit of accuracy in the manufacturing process, and the device-specific characteristics such as lens distortion are present. Arise. For this reason, in the imaging apparatus, eigenvalues such as lens distortion are stored in a memory or the like as data referred to when the control apparatus corrects pixel data. The control device performs a process of reading the eigenvalue from the memory of the imaging device when the work robot is activated. In addition to lens distortion, various data are set as eigenvalues. For example, in an imaging apparatus, information on a defect point related to an imaging element in which the level of an output signal is excessively higher than that of surrounding pixels may be stored in a memory as an eigenvalue.

JP 2012-89552 A

  Also, with communication cables compliant with the above-mentioned communication standards such as the CameraLink standard, the data transfer rate of pixel data is generally set higher than the data transfer rate of data such as parameters that set the imaging conditions of the imaging device. Yes. The eigenvalue is read out by low-speed communication that transmits the parameters of the imaging condition. On the other hand, when the correction data for each pixel is set for eigenvalues such as lens distortion data, the amount of data increases in proportion to the increase in the number of pixels of the mark camera due to higher resolution. In addition, the control device needs to read a plurality of types of eigenvalues when the work robot is activated. For this reason, the work robot has a problem in that the start-up time is delayed due to the control device reading out the eigenvalue by low-speed communication.

  The present invention has been made in view of the above-described problems, and an object thereof is to provide an image processing system and a working robot that can shorten the activation time.

  An image processing system according to a technique disclosed in the present application includes an imaging unit that obtains pixel data obtained by photoelectrically converting light from a workpiece onto an imaging element using a lens, and at least one device of the lens or the imaging element A correction unit that corrects pixel data based on a unique eigenvalue, a low-speed transmission unit that performs reading and writing communication of imaging parameters of the imaging unit, and an effective data transfer rate is faster than a low-speed transmission unit, High-speed transmission means for transmitting pixel data from the image pickup means to the correction means, and eigenvalue transmission means for transmitting the eigenvalue converted into a data format corresponding to the pixel data from the image pickup means to the correction means by the high-speed transmission means. And comprising.

  In addition, the work robot according to the technique disclosed in the present application operates based on pixel data of a work acquired by the imaging unit of the image processing system. The image processing system includes an imaging unit that obtains pixel data in which light from a workpiece is imaged on an imaging element by a lens and photoelectrically converted by the imaging element, and a pixel based on a unique value unique to at least one of the lens and the imaging element. A correction means for correcting data, a low-speed transmission means for reading and writing parameters for imaging of the imaging means, and an effective data transfer rate is faster than that of the low-speed transmission means. High-speed transmission means for transmitting the pixel data, and eigenvalue transmission means for transmitting the eigenvalue converted into the data format corresponding to the pixel data from the imaging means to the correction means by the high-speed transmission means.

  According to the technique disclosed in the present application, it is possible to provide an image processing system and a working robot that can shorten the activation time.

The schematic diagram for demonstrating the image processing system applied to the mounting apparatus of a present Example. The figure which is the data memorize | stored in the non-volatile memory with which a camera is provided, and shows the correspondence of a test pattern and the data corresponding to it. The figure which shows the pixel data of 1 frame which a camera transfers to an image input board. The figure which shows the operation | movement sequence of the process in which an image input board acquires a test pattern from a camera. The figure which shows the operation | movement sequence of the process in which the conventional image input board acquires a test pattern from a camera. The figure which shows the pixel data of 1 frame which the camera of another example transfers to an image input board.

  Embodiments of the present invention will be described below with reference to the drawings. First, an electronic component mounting apparatus (hereinafter sometimes abbreviated as “mounting apparatus”) will be described as an example of a working robot to which the image processing system of the present application is applied. In the embodiment, a configuration in which the image processing system of the present application is applied to data transmission between an image processing board and a camera (imaging device) included in the control device of the mounting device is illustrated.

(Configuration of mounting device 10)
The mounting device 10 in FIG. 1 is an electronic component mounting device that mounts a plurality of electronic components on a circuit board in an integrated circuit production line, for example. The circuit board is applied with solder at the mounting position of the electronic component by a cream solder printing machine on the production line, and is sequentially transported through the plurality of mounting devices 10 to mount the electronic component. Thereafter, the circuit board on which the electronic components are mounted is transferred to a reflow furnace and soldered to constitute an integrated circuit. The camera 3 is, for example, an FA imaging device incorporated in the mounting device 10 for imaging an electronic component or a circuit board. FIG. 1 shows a part of the mounting device 10.

  As shown in FIG. 1, the mounting apparatus 10 includes an image processing board 2, an image input board 4 connected to the image processing board 2, and a camera 3 connected to the image input board 4 so as to be communicable. Has been. The image processing board 2 includes a CPU 11 that processes pixel data PIXD captured by the camera 3 and a RAM (Random Access Memory) 12 that stores processing data in the image processing of the CPU 11. The image input board 4 is connected to, for example, an expansion slot of the image processing board 2 and is connected to the image processing board 2 via the internal bus 5. The image input board 4 includes a command communication unit 21, a CPU 22, a RAM 25, and an FPGA (Field Programmable Gate Array) 26. On the other hand, the camera 3 includes a command communication unit 31, a CPU 32, a nonvolatile memory 35, an FPGA 36, an image sensor 39, and a lens unit 40. The image input board 4 is communicably connected to the camera 3 via a communication cable 7 and transmits and receives various signals. The FPGA 26 of the image input board 4 and the FPGA 36 of the camera 3 perform communication based on, for example, the CameraLink standard through the communication cable 7. The CameraLink standard here is a serial communication standard for transmitting data for pixel data by LVDS (Low Voltage Differential Signaling). The communication cable 7 conforms to the CameraLink standard whose specification is base configuration, for example, and includes three types of signal lines 7A to 7C for transferring pixel data PIXD, command data CMD, and control signals (trigger signal TRIG, etc.). It is provided in the communication cable 7 of the book.

  The command communication unit 21 included in the image input board 4 performs low-speed serial communication (for example, communication based on the RS232C standard) via the command communication unit 31 of the camera 3 and the signal line 7A of the communication cable 7. The command communication unit 21 receives the command data CMD processed by the CPU 22 based on the control of the image processing board 2, and transmits the input command data CMD to the command communication unit 31 of the camera 3. The command communication unit 31 transfers the received command data CMD to the CPU 32. For example, the CPU 22 instructs the CPU 32 to read out the imaging parameters (such as the output gain setting value of the image sensor 39) stored in the memory of the camera 3 and to write the changed setting value into the memory. Data CMD is transmitted. Further, the CPU 32 performs control to change the output gain of the image sensor 39 in accordance with the set value stored in the memory. Or CPU22 transmits the command data CMD which instruct | indicates starting or a stop of the illuminating device (not shown) with which the camera 3 is provided with respect to CPU32, and changes imaging conditions.

  The lens unit 40 of the camera 3 is configured by a lens, a lens holding member, and the like, and forms an image of light from a work (circuit board, electronic component, etc.) that is a subject on the image sensor 39. The image sensor 39 is an image sensor such as a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor). The imaging element 39 photoelectrically converts the light imaged on the imaging surface by the lens unit 40 and outputs it as an analog imaging signal. The image signal output from the image sensor 39 is converted to pixel data PIXD of a digital signal by an A / D conversion circuit (not shown) and output to the FPGA 36. The FPGA 36 is a programmable logic device. The FPGA 36 performs preprocessing such as gain adjustment on the pixel data PIXD acquired from the A / D conversion circuit in accordance with the trigger signal TRIG from the FPGA 26 of the image input board 4 via the signal line 7B, and the signal of the communication cable 7 The pixel data PIXD as a processing result is transferred to the FPGA 26 by high-speed communication via the line 7C.

(About eigenvalues)
The nonvolatile memory 35 is, for example, a flash memory or a ROM. In addition to a program such as a driver for the CPU 32 to control the image sensor 39 and the lens unit 40, the FPGA 26 of the image input board 4 performs a correction process on the pixel data PIXD. A unique value is stored as data to be referred to when performing. The eigenvalue here is data indicating device-specific characteristics of the lens unit 40 and the image sensor 39, such as lens distortion correction, defect correction, gamma correction, and white balance. The characteristic value for lens distortion correction is data for correcting pixel data PIXD in which a subject (work) is distorted due to an error in the attachment position of the lens of the lens unit 40 or the image sensor 39 or the like. Further, the eigenvalue for defect correction is, for example, data related to a pixel (element) that does not react to brightness and does not output a normal pixel value among a plurality of elements (CMOS) included in the image sensor 39. The gamma correction will be described later.

  The CPU 22 of the image input board 4 performs a process of requesting the camera 3 to transmit the above-described plurality of types of eigenvalues when the mounting apparatus 10 is activated and storing it in the RAM 25. The FPGA 26 performs correction processing such as lens distortion correction and gamma correction on the pixel data PIXD acquired from the camera 3, for example, and transfers the processing result to the image processing board 2 via the internal bus 5. The CPU 11 of the image processing board 2 executes various image processes on the pixel data PIXD transferred from the FPGA 26. In this image processing, for example, processing such as binarization of pixel data PIXD, filtering, and hue extraction, and the presence / absence of a subject and position information are recognized based on a specific shape and mark from the processed pixel data PIXD. Recognition processing is included.

  Here, in the communication cable 7 compliant with the above-described communication standard such as the CameraLink standard, the data transfer rate of the pixel data PIXD having a large data amount is generally set higher than the command data CMD. The command data CMD is transmitted / received at a data transfer rate of several hundred kbps by serial communication conforming to the RS232C standard, for example. Further, the pixel data PIXD is transferred at a data transfer rate of several Gbps by serial communication based on the CameraLink standard using the LVDS technology, for example. For example, the pixel data PIXD is transferred at a data transfer rate of 2 Gbps when the number of TAPs (the number of data lines that can be transferred simultaneously) is 3 TAP and the output frequency is 85 MHz. On the other hand, for example, the correction value for each pixel is necessary for the eigenvalue of lens distortion correction. Therefore, when the number of pixels of the camera 3 is increased due to higher resolution, the amount of data increases proportionally. In the conventional mounting apparatus 10, a process of reading a plurality of eigenvalues stored in the nonvolatile memory 35 of the camera 3 by low-speed data communication for transferring the command data CMD is likely to cause a delay in the startup time. Therefore, the mounting apparatus 10 of the present embodiment shortens the startup time by including the eigenvalue in the pixel data PIXD and transmitting it by high-speed data communication.

  More specifically, a test pattern including the above-described eigenvalue data is stored in the nonvolatile memory 35 of the camera 3. FIG. 2 shows an example of a table TB showing test patterns stored in the nonvolatile memory 35 and data corresponding thereto. In the table TB, data corresponding to the number of each test pattern is set. For example, in the table TB, No. 1 to No. 20 store data used for the normal test of the image sensor 39. The test pattern 1 is set with a test pattern for changing a register value of the image sensor 39. The CPU 32 of the camera 3 searches the table TB of the nonvolatile memory 35 in accordance with the command data CMD from the CPU 22 of the image input board 4 and reads the registration value corresponding to the test pattern 1. The CPU 32 outputs the read registration value to the image sensor 39 and switches the output signal of the image sensor 39 from the actual image to a signal corresponding to the test pattern 1. Pixel data PIXD corresponding to the test pattern 1 output from the image sensor 39 is transferred from the FPGA 36 to the CPU 11 of the image processing board 2 through the FPGA 26 of the image input board 4. Based on the received pixel data PIXD of the test pattern 1, the CPU 11 detects and corrects the vertical direction of the image and the image position shift.

  In the table TB, eigenvalues are associated with the 21st and subsequent test patterns. In the test patterns 21 to 28, eigenvalues for lens distortion correction are set. FIG. 3 shows an example of one frame of pixel data PIXD transmitted from the camera 3 to the image input board 4 via the communication cable 7. In the illustrated example, a test pattern in which an eigenvalue for lens distortion correction is set. Is shown. The pixel data PIXD is pixel data in which 1024 vertical and 1024 horizontal pixels are arranged in one frame. In the pixel data PIXD, one pixel has, for example, 8-bit data. For example, the FPGA 36 transfers one frame of pixel data PIXD to the image input board 4 line by line in response to the trigger signal TRIG from the FPGA 26.

The lens distortion correction eigenvalue data requires, for example, 64 bits (32 bits × 2 directions) per pixel in order to correct the position along the vertical and horizontal directions of the pixel data PIXD. . In this case, the data amount of the eigenvalue is a value of the following equation.
Eigenvalue data amount = 64 bits × 1024 × 1024 = 67108864 bits For example, if the effective value of the data transfer rate of the command data CMD is 115200 bps, the time required for data transfer of the eigenvalue is about 582 seconds (= 67108864/115200). Become. On the other hand, when the effective value of the data transfer rate of the pixel data PIXD is 2040000000 bps, the time required for data transfer of the eigenvalue is about 0.03 seconds (= 67108864/2040000000). Accordingly, when only the data transfer rate is compared, if the eigenvalue is included in the pixel data PIXD and the data transfer is performed, the transfer time can be reduced by about 10 minutes. In addition, since one pixel of the pixel data PIXD is 8 bits, the image input board 4 and the camera 3 have a plurality of frames (for example, for transferring lens distortion correction eigenvalues that require 64 bits of correction data per pixel). , 8 (= 64 bits / 8 bits) pixel data PIXD must be transferred. For this reason, in the table TB shown in FIG. 2, one type of eigenvalue for lens distortion correction is divided into eight test patterns 21 to 28. The image input board 4 transfers the eight test patterns 21 to 28 while changing the number of the test pattern designated for the camera 3 and receives the eigenvalue of the lens distortion correction.

  In the table TB shown in FIG. 2, the eigenvalues for gamma correction are associated with the test pattern 29. The data amount of eigenvalues for gamma correction is smaller than the data amount for lens distortion correction. The FPGA 26 of the image input board 4 performs gamma correction for converting the pixel value of the pixel data PIXD as preprocessing of image processing (such as edge detection processing of the circuit board) in the CPU 11 of the image processing board 2, for example. In the test pattern 29 of the table TB, conversion data (look-up table) in which an output pixel value converted to an input pixel value is associated is set. In this lookup table, a value for converting the bit width (gradation) of the pixel value is set. For example, the FPGA 26 converts each pixel value of the pixel data PIXD from 8 bits to a predetermined bit (for example, 6 bits) using a look-up table acquired from the camera 3 and performs bit compression. The table TB may have a configuration in which a plurality of lookup tables having different bit widths after conversion are set in association with each of a plurality of test patterns.

As the data of the lookup table, which is an eigenvalue, for example, 6-bit (64 gradation) data after conversion corresponding to each value (256 gradations) that can be expressed in 8 bits is required. In this case, the data amount of the eigenvalue is a value of the following equation.
Eigenvalue data amount = 256 × 6 bits = 1536 bits In this case, the pixel data PIXD corresponding to the test pattern 29 can assign all the data in the lookup table to the pixel values in the region up to the middle of the first line. It becomes. For this reason, for example, the FPGA 36 of the camera 3 does not target all the pixels of the pixel data PIXD of the test pattern 29 but performs a partial reading process for reading out some pixels corresponding to the eigenvalues. The partial reading process referred to here is, for example, called ROI (Region Of Interest) reading, and is performed by limiting the area to be processed of pixel data. When the test pattern 29 is requested, for example, when reading the pixel data PIXD including the lookup table from the nonvolatile memory 35, the FPGA 36 ends the reading of the pixel data PIXD at the time of reading one line. Only the data for one line is transferred to the image input board 4. Thereby, the FPGA 36 can transfer the lookup table to the image input board 4 at a higher speed.

  Note that, in the test patterns after the 30th in the table TB, pixel data PIXD in which a unique value (defect correction, white balance, etc.) is set as the pixel value is set similarly to the above-described unique value. For example, the eigenvalue for defect correction is normal when a pixel that does not respond to brightness and does not output a normal pixel value or exceeds a predetermined gain value among a plurality of elements (for example, CMOS) of the image sensor 39. Information on pixels that do not output pixel values is set. For each pixel of the pixel data PIXD in which a unique value for defect correction is set, defect point information is set in association with a value expressed by, for example, 8 bits. For example, the defect point information is normal when the 8-bit bit value is “0” and a normal pixel value is not output, and when the bit value is “1”, it is normal when the gain value exceeds level 1 If a bit value is “2” for a pixel that does not output a correct pixel value, a pixel that does not output a normal pixel value when a gain value of level 2 or higher is exceeded, or a level of 3 or higher if a bit value is “3” A bit value corresponding to a gain value that can output a normal pixel value is set.

  Next, a test pattern acquisition process related to the eigenvalues of the image input board 4 will be described. FIG. 4 shows an operation sequence of processing in which the image input board 4 requests a test pattern from the camera 3 and acquires a unique value. In addition, the arrow shown with the continuous line of FIG. 4 has shown serial communication based on RS232C which transfers command data CMD, for example (refer signal line 7A of FIG. 1). Moreover, the arrow shown with a dashed-dotted line shows the communication which transfers trigger signal TRIG by one (for example, CC1) among four types of camera control signals (CC1, CC2, CC3, CC4) prescribed | regulated by CameraLink standard, for example. (See signal line 7B in FIG. 1). Moreover, the arrow shown with a broken line has shown the communication based on the CameraLink specification of the base configuration which transfers pixel data PIXD, for example (refer signal line 7C of FIG. 1). First, the CPU 22 of the image input board 4 sends command data CMD (“command CMD1” in the figure) requesting transmission of a test pattern corresponding to the eigenvalue from the command communication unit 21 to the camera 3 as the mounting apparatus 10 is activated. It transmits to the command communication part 31. For example, the CPU 22 sequentially requests eigenvalues necessary for image processing at the time of activation in accordance with a control program stored in a nonvolatile memory (not shown) provided in the image input board 4. The command CMD1 is, for example, command data CMD requesting the test pattern 21 having a unique value for lens distortion correction. When receiving the command CMD1 from the command communication unit 31, the CPU 32 of the camera 3 transmits a response signal ACK1 indicating that the command CMD1 has been received from the command communication unit 31 to the command communication unit 21 of the image input board 4. The response signal ACK1 is command data CMD directed from the camera 3 to the image input board 4. When the response signal ACK1 from the command communication unit 21 is input, the CPU 22 controls the FPGA 26 to transmit a trigger signal TRIG1 instructing the start of transmission of the test pattern 21 to the FPGA 36 of the camera 3.

  When receiving the trigger signal TRIG1, the FPGA 36 of the camera 3 executes a process of transferring pixel data PIXD including the eigenvalue corresponding to the test pattern 21 stored in the nonvolatile memory 35 to the image input board 4. The FPGA 36 continuously transmits pixel data PIXD corresponding to the test pattern 21 to the FPGA 26 line by line. The FPGA 26 of the image input board 4 stores in the RAM 25 pixel data PIXD for each line continuously transferred from the FPGA 36. The FPGA 26 receives up to the 1024th line and notifies the CPU 22 that reception of the pixel data PIXD of one frame has been completed. The CPU 22 transmits to the camera 3 command data CMD (“command CMD2” in the drawing) for requesting transmission of the next test pattern (in this case, the test pattern 22) corresponding to the eigenvalue of lens distortion correction. When the CPU 22 receives the response signal ACK2 to the command CMD2 from the CPU 32 of the camera 3, the CPU 22 controls the FPGA 26 to transmit a trigger signal TRIG2 for instructing the start of transmission of the test pattern 22 to the FPGA 36 of the camera 3. In this manner, the CPU 22 and the FPGA 26 receive the lens distortion correction eigenvalues included in the test patterns 21 to 28 from the camera 3, store them in the RAM 25, and use them for subsequent processing. When the CPU 22 receives the pixel data PIXD of the test patterns 21 to 28, the CPU 22 requests the test pattern 29 corresponding to the next eigenvalue (for example, gamma correction) and acquires it according to the same operation sequence.

As described above, according to the above-described embodiment, the following effects can be obtained.
<Effect 1> Communication between the image input board 4 and the camera 3 via the communication cable 7 is performed by comparing the camera 3 with the command data CMD (command CMD1 or the like) transmitted and received between the CPU 22 of the image input board 4 and the CPU 32 of the camera 3. The data transfer rate of the pixel data PIXD transferred from to the image input board 4 is set high. In addition, the nonvolatile memory 35 provided in the camera 3 stores a test pattern in which data of eigenvalues unique to the camera 3 such as lens distortion correction are set as pixel values. The CPU 22 transmits to the camera 3 a command CMD1 for requesting transmission of a test pattern corresponding to a necessary eigenvalue when the mounting apparatus 10 is activated (see FIG. 4). Further, when the CPU 22 receives the response signal ACK1 to the command CMD1 from the camera 3, the CPU 22 controls the FPGA 26 to transmit the trigger signal TRIG1 instructing the start of transmission of the test pattern to the FPGA 36 of the camera 3. Upon receiving the trigger signal TRIG1, the FPGA 36 transmits pixel data PIXD corresponding to the requested test pattern. As a result, the mounting apparatus 10 can reduce the processing time for reading the eigenvalue by transferring the pixel data PIXD including the eigenvalue through high-speed communication with a high data transfer rate, and can shorten the activation time. .

  Here, FIG. 5 shows an operation sequence of the acquisition process of eigenvalue data in the prior art. In the conventional example shown in FIG. 5, eigenvalue transfer is processed by low-speed communication (for example, communication conforming to the RS232C standard) for transmitting command data CMD. In general, communication for transferring command data CMD used for reading processing of imaging parameters has a data amount (packet length) in one transfer compared to communication for transferring pixel data PIXD having a large data amount. It is preferable to set a small value. In addition, in the communication of the command data CMD, the amount of data in one transfer may be smaller than the communication of the pixel data PIXD by adding control information such as a header in the command data CMD. Then, as shown in FIG. 5, in the process of transferring the eigenvalue using the communication method applied to the command data CMD, one type of eigenvalue is received in multiple times using a packet with a small amount of data, and A response confirmation for transmitting command data CMD (command CMD1 to command CMD4 in the figure) for requesting the next data every time it is received is executed. On the other hand, as shown in FIG. 4, in the communication in which the eigenvalue is transferred using the pixel data PIXD of the present embodiment, when the camera 3 receives the trigger signal TRIG1, the frame data of the pixel data PIXD is not performed without confirming the response. Are continuously transmitted for each line. Therefore, in the mounting apparatus 10 of the present embodiment, the eigenvalue data can be continuously transmitted, so that the activation time can be shortened.

  Further, the mounting device 10 can shorten not only the startup time but also the time until the work is started after the camera 3 is replaced. More specifically, the eigenvalue such as lens distortion correction is a device-specific characteristic of the lens unit 40 and the image sensor 39. Assume that the mounting apparatus 10 of the present embodiment has a configuration in which the camera 3 is provided on a mounting head that holds an electronic component, and the mounting head is detachable. In this case, when the mounting apparatus 10 is replaced with a different type of mounting head, the image input board 4 needs to read the unique value of the camera 3 included in the replaced mounting head. For this reason, even when the camera 3 is changed by exchanging the mounting head, the mounting apparatus 10 reads the eigenvalue using the pixel data PIXD, thereby reducing the time until the start of work after the replacement.

  <Effect 2> The nonvolatile memory 35 of the camera 3 stores pixel data PIXD including test pattern numbers and eigenvalue data corresponding thereto (see FIG. 2). The CPU 22 of the image input board 4 acquires a unique value by designating a test pattern number by command data CMD. For this reason, the image input board 4 can read out the necessary eigenvalues only by changing only the number of the test pattern to be requested, the processing content for reading out the eigenvalues is simplified, and the startup time can be further shortened. It becomes possible.

  <Effect 3> In the pixel data PIXD corresponding to the test pattern 29, all the data in the lookup table (eigenvalue of gamma correction) are assigned to the pixel values in the region up to the middle of the first line of one frame. The FPGA 36 of the camera 3 does not target all the pixels of the pixel data PIXD of the test pattern 29 stored in the nonvolatile memory 35, but performs a partial reading process for reading out some pixels corresponding to the eigenvalues. Thereby, the FPGA 36 can transfer only necessary data without transferring all the pixel values of the pixel data PIXD in which the unique value is set. Therefore, the mounting apparatus 10 can shorten the activation time because the processing time for reading the eigenvalue is shortened.

In addition, this invention is not limited to the said Example, It is possible to implement in the various aspect which gave various change and improvement based on the knowledge of those skilled in the art.
For example, the communication standard (CameraLink standard) applied to the communication cable 7 is an example, and it is changed to a communication standard based on the CoaXPress standard for transmitting pixel data PIXD at a higher speed than other command data CMD. May be. Further, the image processing system in the present application can be applied to a system in which the effective speed of the data transfer rate between the pixel data PIXD and the command data CMD is different. For example, even if the image processing system is a communication standard in which the pixel data PIXD and the command data CMD are transmitted and received at the same data transfer rate, the communication protocols of the pixel data PIXD and the command data CMD are different from each other. Applicable. Specifically, the present invention can be applied to an image processing system using a communication protocol in which the number of response confirmation processes between the transmission side and the reception side in the data transfer of the pixel data PIXD is smaller than the number of command data CMD processes. In this image processing system, the response confirmation process is simplified, and the eigenvalue is read by communication for transferring the pixel data PIXD in which the effective speed of the data transfer rate is increased even when the same amount of data is transmitted. Thereby, the time required for reading the eigenvalue is shortened.

  In the above embodiment, an error correction detection or error correction code may be added to the test pattern in which the eigenvalue is set. FIG. 6 shows, for example, pixel data PIXD in which an error detection code is added to the test pattern 29 (eigenvalue of gamma correction) of the above embodiment. In the pixel data PIXD shown in FIG. 6, the error detection code CRC is set to the pixel value of the pixels continuous in the region up to the middle of the first line where the gamma correction eigenvalue is set. For example, if the FPGA 26 of the image input board 4 detects an error in the transferred gamma correction eigenvalue (lookup table) data, it notifies the CPU 22 of the error. The CPU 22 performs processing for requesting the camera 3 to retransmit the test pattern. In such a configuration, even when there is no error detection procedure in the communication standard applied to the data transfer of the pixel data PIXD, an image can be obtained by adding an error detection code to each pixel data PIXD in which eigenvalues are set. It is possible to detect errors in eigenvalues on the input board 4 side. Note that the error detection code CRC may be set to another eigenvalue. In addition to error detection, an error correction code (such as FEC) that can correct an error on the receiving side may be added to the lookup table.

Further, the non-volatile memory 35 is stored in a state where the eigenvalue is set in advance to the pixel value of the pixel data PIXD of the test pattern, but may be configured to store only the eigenvalue, for example. In this case, for example, every time the command data CMD specifying the eigenvalue is received, the CPU 32 generates the pixel data PIXD in which the eigenvalue is set to the pixel value and transfers the pixel data PIXD to the FPGA 36. As a result, the storage area of the nonvolatile memory 35 can be reduced, and the manufacturing cost of the camera 3 can be reduced.
A plurality of types of eigenvalues may be set in one frame of pixel data PIXD stored in the nonvolatile memory 35. For example, defect correction and white balance eigenvalues may be set for the pixel values of the second and subsequent lines of the pixel data PIXD shown in FIG.

  In addition, the image input board 4 and the camera 3 transmit and receive the pixel data PIXD and the command data CMD through one communication cable 7, but each data may be transmitted through different communication cables.

  Moreover, although the said Example demonstrated the electronic component mounting apparatus 10 which mounts an electronic component on a circuit board, this application is not limited to this, It applies to the working robot etc. which operate | move in other various production lines can do. For example, the present invention may be applied to a working robot that performs assembly work such as a secondary battery (such as a solar battery or a fuel cell). The working robot includes a screen printing apparatus that moves the squeegee along the mask and prints the printing material on the print target member. Further, the work robot is not limited to a machine that performs mounting and assembly, but includes a machine tool that performs other work such as cutting.

The correspondence with the terms in the claims is as follows.
The electronic component mounting apparatus 10 is an example of a work robot. The camera 3 is an example of an imaging unit. The image input board 4 including the FPGA 26 is an example of a correction unit. The lens unit 40 is an example of a device including a lens. Serial communication conforming to the RS232C standard in which the command communication units 21 and 31 transmit command data CMD constitutes a low-speed transmission unit. Serial communication conforming to the CameraLink standard in which the FPGAs 26 and 36 transmit the pixel data PIXD constitutes high-speed transmission means. The pixel data PIXD in which the unique value stored in the nonvolatile memory 35 is set as the pixel value is an example of the unique value converted into a data format corresponding to the pixel data. The FPGA 36 of the camera 3 that transmits pixel data PIXD including eigenvalues in response to trigger signals TRIG1 and TRIG2 from the FPGA 26 is an example of eigenvalue transmission means. The circuit board and the electronic component mounted on the circuit board are an example of a workpiece.

  10 Electronic component mounting device, 3 camera, 4 image input board, 35 nonvolatile memory, CMD command data, PIXD pixel data, ACK1, ACK2 response signal, TRIG1, TRIG2 trigger signal.

Claims (5)

Imaging means for imaging light from a workpiece onto an image sensor with a lens and obtaining pixel data photoelectrically converted by the image sensor;
Correction means for correcting the pixel data based on a characteristic value unique to at least one of the lens and the image sensor;
Low-speed transmission means for performing reading and writing communication of imaging parameters of the imaging means;
A high-speed transmission means in which an effective speed of a data transfer rate is faster than the low-speed transmission means, and the pixel data is transmitted from the imaging means to the correction means;
Eigenvalue transmission means for causing the high-speed transmission means to transmit the eigenvalue converted into a data format corresponding to the pixel data from the imaging means to the correction means;
An image processing system comprising:   The image processing system according to claim 1, wherein the eigenvalue transmission unit associates the eigenvalue with a test pattern for testing the image sensor and transmits the test value using the high-speed transmission unit.   3. The image processing system according to claim 1, wherein the eigenvalue transmission unit assigns the eigenvalue to a pixel value in a predetermined region in one frame of data corresponding to the pixel data.   4. The image processing system according to claim 1, wherein the eigenvalue transmission unit adds an error code for detecting an error to the data after conversion of the eigenvalue.   5. A working robot having the image processing system according to claim 1 and operating based on the pixel data of the workpiece acquired by the imaging means.

Images