A New Machine Vision System without Programming

Machine vision is the use of machines instead of human eyes to make measurements and judgments. It can quickly acquire a large amount of information, and it is easy to handle automatically. It is also easy to integrate with design information and processing control information. Therefore, in the modern automated production process, people will The vision system is widely used in the fields of condition monitoring, product inspection and quality control. Because of its high precision, non-contact, stability and other characteristics, it has been widely used in the domestic and foreign industrial fields, greatly improving the product quality and automation of the production line.

In general, there is still a gap between domestic research and development of machine vision products and foreign countries. At present, the level of research and development of machine vision products is still at a relatively low level. Commonly used foreign products. At present, applications are mainly focused on detection, positioning, motion control, and robotics. In the past two years, the application requirements for machine vision have grown most rapidly in several areas such as motion control, production lines, diagnosis, testing, and numerical control equipment. Research on machine vision software is also deepening. With more and more applications in the industry, the requirements are getting higher and higher. Currently, machine vision systems on the market have been more difficult to meet the needs of the industry, and there are widespread drawbacks such as difficult to develop, difficult to use, and difficult to sell. These shortcomings are mainly manifested in the fact that the standardization degree of the system is not high enough, and every customer needs to invest a large amount of development manpower, and the developed system has weak reusability, poor user experience, and difficult maintenance and high cost in the later period. Therefore, it is necessary to develop a machine vision system that requires no programming, simple configuration, hardware and software configuration, and simple sales.

This article introduces a new type of machine vision system that does not require programming. It addresses the above-mentioned market conditions and has the following advantages: 1. It can be configured, requires no programming, and is as simple and easy to use as household appliances; 2. It provides a universal application platform. Contains numerous units of machine vision tools; 3, is not based on a smart camera or general vision system development of the application system, it integrates the bottom development and application development in one; 4, carried out a vertical integration, the core system and application process Integration; 5, modularization, providing support for equipment manufacturers, end-users after loading; 6, system sales model simplification; 7, system cost structure optimization, is a product with obvious price advantage. The following describes the detailed design ideas of the system from two aspects of hardware and software.

First, the hardware framework

The system is based on image analysis technology to monitor and measure the operation process of the equipment. Using features extraction and template comparison, the image collected in real time from the scene is compared with the standard image of normal working conditions to achieve abnormality, analysis, judgment and alarm. Shutdown function. The standard template is based on the analysis of images under normal conditions, and the features are extracted. When comparing, the features (combination) of the template are compared with the corresponding features in the real-time image of the scene to determine whether any abnormality has occurred. This monitoring method does not replace the self-diagnosis function of automatic equipment, but it has a good effect in preventing the deviation of the operation process, preventing the production of defective products, and even destroying the equipment. The product consists of the following major components:

1. DSP/ARM-based core processor

2. Optical lens, image sensor, front-end processing and interface (ie camera part)

3.FPGA acts as an accelerator and peripheral interface logic for image processing

4.GPIO and serial communication

5. Ethernet communication

Among them DSP/ARM processor adopts TI OMAP4460 CPU, its ARM Cortex-A9 main frequency reaches 1.8GHz, built-in DSP, this is higher than the performance level of the hardware platform of most intelligent cameras so far; The resolution of the camera is mainly 1280x1024, at the same time Supports cameras of 5 megapixels or more; memory supports NOR FLASH and NAND FLASH, and can also expand the SD card; FPGA as a logic expansion device for peripherals, mainly by two roles: responsible for image data buffering and preprocessing , and then sent to the DSP processor for the corresponding algorithm processing; communications support Ethernet and serial port, can be easily connected with the robot, PLC and other devices; I / O settings with GPIO (general input and output), the connection to send the trigger over the scene Signals (inputs, triggered photographs), and control signals (outputs) sent to the device.

In addition, the system can provide two types of products:

Visual tablet. A hardware platform with various functional interfaces for vision applications, featuring the following features: LCD monitor and touch screen function (large screen); built-in high-performance processor (ARM architecture); camera access; I/O interface: RS-232 , USB, 10/100/1000 Ethernet; Network Communication Function: WiFi; Encryption protection.

Second, software technology

The ARM processor in the system is mainly responsible for resource management and logic control, equipped with Angstrom Linux system, and integrates the drivers for the functional modules required by machine vision. The specific contents of the system software development include: USB camera, GPIO, LCD touch screen, RS-232 driver, Angstrom Linux system porting. Angstrom Linux is a distribution desktop system running on an embedded platform. It integrates most of the system tools and libraries, including the QT and opencv libraries for software development and algorithm development, and can be selected via online compilation. The tool library you need is highly customizable and ideal for use in resource-poor embedded products. The driver of the underlying module is modified by the linux official kernel.

The application layer software development is based on the Ubuntu system, and QT is used for interface design and logic design of the software. Ubuntu is a free, open-source desktop PC operating system based on Linux that can be easily configured in similar environments to embedded platforms such as QT and opencv. The software design goal of the system is no programming, simple configuration, modularization, so all possible visual tools are integrated into an integrated software platform vdStudio, including common tools such as positioning, measurement, detection, calibration, and so on. With a graphical user interface, users can easily select and configure each tool, and the interface is designed based on the user's habits. It is easy to learn and use. In addition, the system supports offline operation, one-time configuration, and permanent operation. As shown in Figure 4.

Third, algorithm research

The algorithm is the core of the machine vision system. Compared with other parts of the system, the algorithm contributes the most to the visual function. From the perspective of market share, the sales of standard machine vision systems based on core software algorithms account for almost half of the entire machine vision industry, and the importance of algorithms and software can be seen. If the algorithm is not good, the visual function can not be achieved, or although it can achieve the function, but the performance is not competitive. The algorithm is very good, but the software is not well-written will directly affect the visual system function optimization and performance indicators. Algorithms and software systems are the most important accumulation of visual companies that use standard systems as their products. In the market competition, whether the algorithm is competitive or not plays a very key role. In addition to the existing accumulation, how to obtain more algorithm software resources for my own use is also one of the key factors that determine the success or failure of this project. In addition to developing its own algorithms in some subdivided areas, we can use internationally outstanding software and algorithms to find and use them as an important path to product improvement through cooperation and sharing models.

The system software design uses a variety of excellent design patterns, making it possible to flexibly choose which algorithm to use when it is used. Among them, the independently developed VD300 SDK (software development kit) includes algorithm functions such as CG algorithm, basal plane correction, image splicing, and scanning control; the algorithm SDK developed based on opencv includes tools for template matching, measurement, and blob detection; The machine vision function library IPT (Image Processing Toolset) and so on.

Fourth, the conclusion

The system can be used directly by visual appliances, does not require ASP integration, does not require programming, like household appliances, by reading the user manual can easily get started. From the manufacturer to the user, there is no technical integration link, and the saved cost can also benefit the user and reduce the product price. The users get the required functions, do not need to understand too many technical details, just identify the target users when selling, explain the function of the product, and stare at the competitive advantage, the ordinary sales clerk does not need too strong technical background to be competent.