Machine vision measurement system and method for adjusting simulation parameters of sensor chip

Abstract

Disclosed is a machine vision measurement system, which comprises an image sensor chip and a control processor for cooperative use. The control processor further comprises a high dynamic range imaging system module. The high dynamic range imaging system module comprises: a data receiving module, which is used for synchronizing a row starting point and ending point on the basis of the number of row data; a data feature value analysis module, which is used for carrying out feature value extraction on serial data; a parameter decision feedback module, which determines whether the extracted feature value is located in a preset optimal working area, keeping simulation parameters of an image sensor chip unchanged if so, and if not, adjusting the simulation parameters of the image sensor chip; a synchronous system control generator module, which is used for tracking a control working time sequence of the image sensor chip and starting up a data input detection window; and a synchronous simulation parameter output module, which updates the simulation parameters of the image sensor chip on the basis of a signal of the synchronous system control generator module.

Description

Machine vision measurement system and sensor chip simulation parameter adjustment method

[0001] This application claims priority to Chinese Patent Application No. 202311586031.5 filed on November 24, 2023, which is incorporated by reference in its entirety.

Technical field

[0002] The present invention belongs to the field of measurement, and in particular relates to a machine vision measurement system for achieving intra-frame high dynamic range imaging through dynamic feedback and a sensor chip simulation parameter adjustment method. [Background Technology]

[0003] High Dynamic Range Imaging (HDR) refers to a set of technologies used in computer graphics and cinematography to achieve a greater dynamic range (i.e., greater differences in brightness and darkness) than conventional digital imaging techniques. In industrial machine vision, HDR is often used to address scenes with large differences in brightness or imaging interference caused by multiple reflections.

[0004] Currently known methods include inter-frame comparison and single-frame high dynamic range imaging. The inter-frame comparison method involves capturing two or more different image frames for comparison, then combining them with algorithmic design to make a decision. This method includes two approaches: varying exposure times and gain configurations. This method requires multiple frames of runtime. Furthermore, as the object under test is in motion, the images sampled between frames may not be from the same location on the object, introducing inter-frame interference. The single-frame high dynamic range imaging method, on the other hand, can be implemented using a single frame, primarily through varying exposure times, creating a nonlinear quantization method. This method relies on exposure time rather than actual optical response, so in practice, data processing that requires differentiation is not implemented due to nonlinear conversion nodes. This limits its effectiveness and makes it unable to respond to differences in the sensitivity of the object under test. While single-frame high dynamic range imaging is possible, its actual frame rate is only about half that of the non-high dynamic range imaging mode, with a comparable impact on frame rate as the inter-frame comparison method.

[0005] All of the above methods have a significant impact on the high frame rate sampling requirements of machine vision applications, which will cause the frame rate to be nearly halved, and therefore cannot be applied to the machine vision field.

[0006] [Summary of the invention]

[0007] The object of the present invention is to provide a machine vision measurement system and a sensor chip simulation parameter adjustment method to solve the problem that the prior art has a significant impact on the frame rate.

[0008] To achieve the above objectives, a machine vision measurement system according to the present invention is implemented, comprising an image sensor chip and a control processor for use together. The image sensor chip includes a photosensor array, a sampling readout circuit, an operational amplifier, an analog-to-digital converter, an array exposure controller, a system control generator, and a communication interface configuration module. The control processor includes a communication interface configuration module, a main control system, a data output system, and a data processing system. The control processor is characterized in that it also includes a high dynamic range imaging system module, and the high dynamic range imaging system module includes:

[0009] A data receiving module is used to synchronize the row start and end points according to the number of row data;

[0010] The data eigenvalue analysis module is used to extract eigenvalues ​​from serial data. Specifically, after calibrating the starting point and end point of each line, eigenvalue extraction is performed after receiving each line of data.

[0011] The parameter decision feedback module determines whether the extracted eigenvalue of the current row is within a preset optimal working area. If so, the analog parameters of the image sensor chip remain unchanged. If it is above the optimal area, the analog parameters of the image sensor chip are adjusted in a downward trend of the eigenvalue; otherwise, the analog parameters of the image sensor chip are adjusted in an upward trend of the eigenvalue.

[0012] A synchronous system control generator module is used to track the control operation timing of the image sensor chip and start the data input detection window;

[0013] The synchronous simulation parameter output module updates the simulation parameters of the image sensor chip according to the signal of the synchronous system control generator module.

[0014] According to the above main features, the characteristic value is one of the pixel maximum value, average value or contrast.

[0015] According to the above main features, the simulation parameters include gain parameters and quantization parameters.

[0016] To achieve the above object, the present invention provides a method for adjusting the analog parameters of an image sensor chip using the above machine vision measurement system, the method comprising the following steps:

[0017] Obtain the judgment interval parameters and set the optimal working area of ​​pixel features;

[0018] After the system starts working, the synchronous system control generator module tracks the control working timing of the image sensor chip and starts the data input detection window;

[0019] The data receiving module synchronizes the row start and end points according to the number of row data;

[0020] The data eigenvalue analysis module extracts eigenvalues ​​from the serial data. Specifically, after calibrating the starting point and end point of each line, it extracts eigenvalues ​​after receiving the data of each line.

[0021] The parameter decision feedback module determines whether the extracted eigenvalue of the current row is within the set optimal working area. If so, the simulation parameters remain unchanged. If it is above the optimal area, the simulation parameters of the image sensor chip are adjusted in a downward trend of the eigenvalue. Otherwise, the simulation parameters of the image sensor chip are adjusted in an upward trend of the eigenvalue.

[0022] The synchronous analog parameter output module updates the analog parameters of the image sensor chip according to the synchronous system control signal.

[0023] According to the above main features, the characteristic value is one of the pixel maximum value, average value or contrast.

[0024] According to the above main features, the simulation parameters are gain parameters and quantization parameters.

[0025] According to the above main features, adjusting the analog parameters of the image sensor chip along the upward trend of the eigenvalue is to increase the gain parameter and decrease the quantization parameter, while adjusting the analog parameters of the image sensor chip along the downward trend of the eigenvalue is to decrease the gain parameter and increase the quantization parameter.

[0026] According to the above main features, the analog parameters are gain parameters and quantization parameters. The method for adjusting the analog parameters of the image sensor chip includes the following steps:

[0027] An equivalent gain value table is established according to possible values ​​of the gain parameter and the quantization voltage, wherein the equivalent gain is the ratio of the gain parameter to the quantization voltage;

[0028] Get the eigenvalue V0 and equivalent gain K0 of the current row;

[0029] Determine the amplification factor K', where the amplification factor K'=expected equivalent gain K1 / equivalent gain K0;

[0030] Enumerate the new expected target values ​​of all magnifications, where the new expected target value is the eigenvalue V0 of the current row × the magnification K';

[0031] The amplification factor K' is determined according to the new expected target value falling into the optimal working area, and then the expected equivalent gain K1 is determined. Then, the gain parameter and quantization parameter corresponding to the expected equivalent gain K1 are determined according to the above equivalent gain value table.

[0032] According to the above main features, if there are multiple new expected target values ​​falling into the optimal working area, the new expected target value closest to the middle value of the optimal working area is selected to determine the magnification K'.

[0033] Compared with the existing technology, the present invention uses the conversion time between rows to complete the calculation feedback and feature configuration changes, thereby completing continuous adjustment within a single-frame image and tracking the continuous changes of the image. It has the following technical effects: First, through the single-frame high dynamic range imaging mode, there is no need to reduce the frame rate, and there will be no information difference interference between different frames; second, different positions and scenes in the frame are tracked in real time using different gain configurations and dynamic quantization ranges to prevent the problem of information being too weak or overexposed.

Brief Description of the Drawings

[0034] FIG1 is a schematic diagram showing the functional modules of a machine vision measurement system according to the present invention.

[0035] FIG2 is a schematic diagram of the working process of the machine vision measurement system implementing the present invention.

[0036] FIG3 is a schematic diagram of the workflow of the high dynamic range imaging system module. [Specific implementation method]

[0037] Please refer to FIG1 , which is a schematic diagram of the functional modules of a machine vision measurement system according to the present invention. The machine vision measurement system according to the present invention includes:

[0038] The host computer system is used to receive input commands from the user and send them to the control processor, and receive data output by the control processor and present it to the user;

[0039] The control processor includes a communication interface configuration module, a main control system, a data output system, and a data processing system. The operating principles of the communication interface configuration module, the main control system, the data output system, and the data processing system are described in the prior art and will not be described in detail here. In addition, the control processor also includes a high dynamic range imaging system module;

[0040] The image sensor chip includes a photosensor array, a sampling readout circuit, an operational amplifier, an analog-to-digital converter, an array exposure controller, a system control generator, and a communication interface configuration module. The operating principles and methods of the photosensor array, sampling readout circuit, operational amplifier, analog-to-digital converter, array exposure controller, system control generator, and communication interface configuration module are well known in the prior art and will not be described in detail here.

[0041] The improvement of the present invention lies in that a high dynamic range imaging system module is added to the control processor, and the high dynamic range imaging system module includes:

[0042] The data receiving module is used to synchronize the starting and ending points of a row according to the number of row data and mark the corresponding data valid / invalid flags;

[0043] A data eigenvalue analysis module is used to extract eigenvalues ​​from serial data. The eigenvalues ​​can be a combination of one or more pixel maximum, minimum, mean, contrast, grayscale spot size, etc.

[0044] a parameter decision feedback module, which determines whether the extracted eigenvalue is within a preset optimal working area, and if so, maintains the simulation parameters of the image sensor chip unchanged; if it is above the optimal area, adjusts the simulation parameters of the image sensor chip in a downward trend of the eigenvalue; otherwise, adjusts the simulation parameters of the image sensor chip in an upward trend of the eigenvalue;

[0045] A synchronous system control generator module is used to track the control operation timing of the image sensor chip and start the data input detection window;

[0046] The synchronous analog parameter output module updates the analog parameters of the image sensor chip according to the signal of the synchronous system control generator module. In a specific implementation, the analog parameters include gain parameters and quantization parameters.

[0047] Please refer to FIG2 , which is a schematic diagram of the workflow of the machine vision measurement system according to the present invention. The workflow of the machine vision measurement system according to the present invention includes the following steps:

[0048] The system is powered on, and the image sensor chip and control processor complete the power-on process;

[0049] The control processor completes the recognition of the image sensor chip, and the host computer completes the configuration of the functions and parameters of the control processor and the image sensor chip;

[0050] Control the processor and image sensor chip to enter a standby state, waiting for a start-up instruction;

[0051] The host computer sends a start command to the control processor, which responds to the start command and sends an event trigger signal to the image sensor chip. Each signal triggers the sensor to complete the complete process of exposure and data transmission.

[0052] The image sensor chip generates control signals according to the event trigger signal, including exposure control signal, sampling control signal and data output control signal, and the control processor generates the above signals synchronously;

[0053] Line-by-line exposure control, where the exposure timing implements CDS (correlated double sampling), and the time gap between line operations is used for the time loss of the HDR processing flow, so that the conversion data matches the simulation parameters;

[0054] Pixel voltage sampling, where the sampling and readout circuit supports CDS requirements and completes the sampling of reset voltage and exposure voltage. In each row cycle, all pixels in a row are sampled and read out using the sampling and readout circuit to be output to a conversion circuit consisting of an operational amplifier and an analog-to-digital converter;

[0055] A row of data is serially controlled and output in sequence, then sampled by the operational amplifier, converted by the analog-to-digital converter and output to the high dynamic range imaging system module for processing and updating of analog parameters.

[0056] Please refer to Figure 3, which is a schematic diagram of the workflow of the high dynamic range imaging system module. The workflow of the high dynamic range imaging system module specifically includes the following steps:

[0057] After powering on, the control processor system performs initialization and configuration. The high dynamic range imaging system module acquires the decision interval parameters and sets the optimal operating area for pixel features. It also configures the image sensor chip array's specifications for synchronizing control and data transmission between the image sensor chip and the control processor. This allows the control processor to synchronize timing operations on the image sensor chip and accurately predict the image sensor chip's operating node. The decision interval parameters can be configured and updated via the host computer. When a photoelectric sensor is exposed to light, it generates a photocurrent. Stronger light intensity increases the photocurrent, longer discharge time, and higher response voltage. Ultimately, depending on device parameters, it undergoes a process of linear voltage increase, saturation, and overexposure. The saturation and overexposure intervals represent the nonlinear relationship between voltage change and light intensity over time. These intervals cannot reflect true intensity contrast. The decision parameter interval, based on the device's operating principle and data processing requirements for data feature intervals, generally refers to the data interval where the characteristic value lies between the non-saturated region and the near-linear region. For example, using an 8-bit grayscale bit depth, this interval is roughly between [140,180].

[0058] Enter standby mode;

[0059] After the host computer starts the system, the synchronous system control generator module tracks the control working sequence of the image sensor chip and starts the data input detection window;

[0060] The data receiving module synchronizes the row start and end points according to the number of row data and identifies the corresponding data valid flag / invalid flag;

[0061] The data feature value analysis module extracts feature values ​​from the serial data. That is, after the start and end points of each line are calibrated, feature value extraction is performed after receiving each line of data. The feature value can be one of the pixel maximum value, mean value or contrast. Preferably, the pixel maximum value is used as the feature value.

[0062] The parameter decision feedback module determines whether the extracted eigenvalue is within the set optimal working region. If so, the simulation parameters remain unchanged; if it is above the optimal region, the eigenvalue is adjusted along the downward trend, and vice versa, the eigenvalue is adjusted along the upward trend.

[0063] The synchronous simulation parameter output module updates the simulation parameters of the image sensor chip at an appropriate timing position according to the synchronous system control signal. In a specific implementation, the simulation parameters are gain parameters and quantization parameters.

[0064] The following uses a specific embodiment to illustrate the above implementation process. Among them, simulation parameter 1 is the gain parameter Gain, simulation parameter 2 is the quantization parameter adc_ref, corresponding to the quantization voltage Vref, the exposure voltage is Ve, and the conversion formula of the analog-to-digital converter is the eigenvalue ADC_DATA = exposure voltage Ve × gain parameter Gain / (quantization voltage Vref / 1024) = exposure voltage Ve × gain parameter Gain × 1024 / quantization voltage Vref. First, set the initial working conditions. Let the gain parameter Gain be 1, the quantization voltage be 1V, and the analog-to-digital converter be 10Bit. Then the eigenvalue ADC_DATA = 1024 × exposure voltage Ve / quantization voltage Vref;

[0065] Taking a single value as an example for the decision-making process, assume that the optimal working region is an interval [Dmin, Dmax]. When the eigenvalue ADC_DATA is between Dmin and Dmax, the two simulation parameters remain unchanged; when the eigenvalue ADC_DATA < Dmin, the eigenvalue ADC_DATA can be increased by increasing the gain parameter Gain and decreasing the quantization voltage Vref; when the eigenvalue ADC_DATA > Dmin, the eigenvalue ADC_DATA can be decreased by decreasing the gain parameter Gain and increasing the quantization voltage Vref.

[0066] Due to design constraints, the gain parameter Gain and the quantization voltage Vref cannot be completely continuous and can be represented as arbitrary values. For simplicity, a lookup table can also be used to determine their values, such as the formula above: Eigenvalue ADC_DATA = Exposure voltage Ve × Gain parameter Gain / (Quantization voltage Vref / 1024) = Exposure voltage Ve × Gain parameter Gain × 1024 / Quantization voltage Vref = Exposure voltage Ve × 1024 × Gain parameter Gain / Quantization voltage Vref. Gain parameter Gain / Quantization voltage Vref = Equivalent gain K. Thus, eigenvalue ADC_DATA = Exposure voltage Ve × 1024 × Equivalent gain K. In actual applications, the gain parameter Gain has a greater impact on the eigenvalue ADC_DATA. In specific designs, the gain parameter Gain can be set to 1, 2, or 4, while the quantization voltage Vref can be set to 0.8, 1, 1.2, or 1.4. Using the formula Equivalent gain K = Gain parameter Gain / Quantization voltage Vref, a value table for the equivalent gain K can be established, as shown in Table 1. Admittedly, the values ​​of the gain parameter Gain and the quantization voltage Vref in the table are only examples. The values ​​of the gain parameter Gain and the quantization voltage Vref can be more precise, so that the equivalent gain K has more values, thereby making the adjustment process more accurate.

[0067] Table 1

[0068] After establishing the value table of the equivalent gain K, the following steps are used to update the simulation parameters:

[0069] The data eigenvalue analysis module extracts eigenvalues ​​from the serial data, that is, after receiving each row of data, it extracts the eigenvalue V0 of the current row;

[0070] If the eigenvalue of the current row is within the set optimal working area, the simulation parameters are kept unchanged; otherwise, the current equivalent gain K0 is obtained, where the equivalent gain K0 = gain parameter Gain / quantization voltage Vref;

[0071] Then, use Table 2 to enumerate the new expected target values ​​of all amplification factors. After the equivalent gain K0 is determined, the expected equivalent gain K1 can be determined according to Table 2. As shown in Table 1, the value ranges of the equivalent gain K0 and the equivalent gain K1 are the same. Therefore, after the equivalent gain K0 is determined, 12 possible values ​​of the expected equivalent gain K1 can be obtained through Table 2. If the equivalent gain K0 = 1.25, then the expected equivalent gain K1 can be taken as: 1.25, 1, 0.833, 0.714, 2.5, 2, 1.66667, 1.4286, 5, 4, 3.33, 2.857, that is, a total of 12 values; then calculate the amplification factor K' = expected equivalent gain K1 / equivalent gain K0; then obtain the new expected target value = the eigenvalue V0 of the current row × amplification factor K';

[0072] Table 2

[0073] The amplification factor K' is determined based on the new expected target value falling within the optimal working area, and then the expected equivalent gain K1 is determined. Then, the gain parameter Gain and the quantization parameter adc_ref corresponding to the expected equivalent gain K1 are determined according to Table 1, thereby updating the above two analog parameters of the image sensor chip.

[0074] In the previous step, if there are multiple new expected target values ​​that fall within the optimal working region, the new expected target value that is closest to the middle value of the optimal working region may be selected.

[0075] Compared with the existing technology, the present invention uses the conversion time between rows to complete the calculation feedback and feature configuration changes, thereby completing continuous adjustment within a single-frame image and tracking the continuous changes of the image, which has the following technical effects: First, through the single-frame HDR mode, there is no need to reduce the frame rate, and there will be no information difference interference between different frames; second, different positions and scenes in the frame are tracked in real time using different gain configurations and dynamic quantization ranges to prevent the problem of too weak information or overexposure.

[0076] It is understandable that those skilled in the art can make equivalent substitutions or changes based on the technical solution and inventive concept of the present invention, and all these changes or substitutions should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A machine vision measurement system, comprising an image sensor chip and a control processor for use together, wherein the image sensor chip comprises a photoelectric sensor array, a sampling readout circuit, an operational amplifier, an analog-to-digital converter, an array exposure controller, a system control generator and a communication interface configuration module, and the control processor comprises a communication interface configuration module, a main control system, a data output system and a data processing system, characterized in that The control processor also includes a high dynamic range imaging system module, and the high dynamic range imaging system module includes: A data receiving module is used to synchronize the start and end points of a row according to the number of row data; The data feature value analysis module is used to extract the feature value of the serial data. Specifically, after the start and end points of each line are calibrated, the feature value is extracted after receiving the data of each line. The parameter decision feedback module determines whether the extracted characteristic value of the current row is located in a preset optimal working area. If so, the simulation parameters of the image sensor chip are kept unchanged; if it is located above the optimal area, the simulation parameters of the image sensor chip are adjusted along the downward trend of the characteristic value, otherwise, the simulation parameters of the image sensor chip are adjusted along the upward trend of the characteristic value; A synchronous system control generator module is used to track the control operation timing of the image sensor chip and start the data input detection window; The synchronous simulation parameter output module updates the simulation parameters of the image sensor chip according to the signal of the synchronous system control generator module.

2. The machine vision measurement system according to claim 1, characterized in that: The characteristic value is one of a pixel maximum value, a pixel mean value or a pixel contrast.

3. The machine vision measurement system according to claim 2, characterized in that: The simulation parameters include a gain parameter and a quantization parameter.

4. A method for adjusting the analog parameters of an image sensor chip using the machine vision measurement system of claim 1, comprising the following steps: Obtain the judgment interval parameters and set the optimal working area of ​​the pixel features; After the system starts working, the synchronous system control generator module tracks the control working timing of the image sensor chip and starts the data input detection window; The data receiving module synchronizes the row start and end points according to the number of row data; The data feature value analysis module extracts feature values ​​from the serial data. Specifically, after calibrating the starting point and the end point of each line, feature value extraction is performed after receiving the data of each line. The parameter decision feedback module determines whether the extracted characteristic value of the current row is located in the set optimal working area. If so, the simulation parameters are kept unchanged; if it is located above the optimal area, the characteristic value is adjusted along the downward trend of the simulation parameters of the image sensor chip, otherwise, the characteristic value is adjusted along the upward trend of the simulation parameters of the image sensor chip; The synchronous analog parameter output module updates the analog parameters of the image sensor chip according to the synchronous system control signal.

5. The method according to claim 4, characterized in that: The characteristic value is one of a pixel maximum value, a pixel mean value or a pixel contrast.

6. The method according to claim 4, characterized in that: The simulation parameters are a gain parameter and a quantization parameter.

7. The method according to claim 4, characterized in that: The analog parameters of the image sensor chip are adjusted along the upward trend of the characteristic value by increasing the gain parameter and decreasing the quantization parameter, while the analog parameters of the image sensor chip are adjusted along the downward trend of the characteristic value by decreasing the gain parameter and increasing the quantization parameter.

8. The method according to claim 4, characterized in that: The simulation parameters are gain parameters and quantization parameters. The method for adjusting the simulation parameters of the image sensor chip comprises the following steps: An equivalent gain value table is established according to possible values ​​of the gain parameter and the quantization voltage, wherein the equivalent gain is the ratio of the gain parameter to the quantization voltage; Get the eigenvalue V0 and equivalent gain K0 of the current row; Determine the amplification factor K', where the amplification factor K'=expected equivalent gain K1 / equivalent gain K0; Enumerate all new expected target values ​​of magnifications, where the new expected target value is the feature value V0 of the current row × magnification K'; The amplification factor K' is determined according to the new expected target value falling into the optimal working area, and then the expected equivalent gain K1 is determined, and then the gain parameter and quantization parameter corresponding to the expected equivalent gain K1 are determined according to the above equivalent gain value table.

9. The method according to claim 8, characterized in that: If there are multiple new expected target values ​​falling within the optimal working region, the new expected target value closest to the middle value of the optimal working region is selected to determine the magnification K'.

Images