At present, machine vision technology is mainly used in the four main directions of positioning, measurement, detection and recognition in China. In the process of image acquisition and image processing, there are several technical difficulties, which directly affect the accuracy of the acquisition of analysis data. Let's take a look at the technical difficulties in the design of machine vision system.
First, calibration
Calibration is generally required in high-precision measurement, which is mainly divided into optical distortion calibration, projection distortion calibration, object and image space calibration, etc. However, the general calibration algorithm is based on plane calibration, and it is difficult to use calibration algorithm to solve the problem of non-plane calibration. There are also some special measurements in which the calibration plate is not used, so the calibration algorithm may not solve all problems.
Second, the measurement accuracy of the software
The measurement accuracy is generally 1/2-1/4 pixels, because the measurement software accuracy is too small, the less feature point information extracted from the image.
Through the above five problems, it can be seen that there will be many factors of interference in the design of machine vision system, so we need to constantly improve and upgrade technology to avoid these inevitable problems, but also believe that technological innovation, many problems can not be solved.
Third, the speed of motion of the object
In the process of image acquisition, image blur accuracy depends on the object motion speed and camera exposure time, so the fast object motion speed is likely to lead to image blur, in addition, other equipment (industrial cameras, industrial lenses, light sources, etc.) will also indirectly affect the detection.
Fourth, the consistency of the workpiece location
The inconsistency of the workpiece position is a particularly important element of concern leading to measurement inspection. Whether it is online detection or offline detection on the industrial production line, the machine vision system needs to ensure that the position reached by the product is the same position when obtaining the product picture information, otherwise the positioning will be inaccurate, the position of the measurement tool will be inaccurate, resulting in the deviation of the measurement results, and it is difficult to control the qualification of the product.
Fifth, the stability of lighting
In machine vision applications, lighting stability affects the accuracy of measurement the most, because as long as the illumination changes slightly, the measurement results may be 1 to 2 pixels wrong. The main reason is that the instability of illumination will affect the change of the edge position of image acquisition, so the influence of ambient light should be eliminated to the minimum in the design of machine vision system, and the stability of the supporting active light source should be ensured.
Through the above, it is not difficult to see that there will be many factors interference in the design of machine vision system, so we need to constantly improve and upgrade the technology to avoid these inevitable problems