"Visual Inspection Solution" Development Trend of Machine Vision Inspection Technology

1. Image acquisition technology develops rapidly

   CCD, CMOS and other firmware are becoming more and more mature, the size of image-sensitive devices continues to shrink, the number of pixels and data rates continue to increase, the speed of resolution and frame rate improvement can be said to be rapid, and the product series is becoming more and more abundant. Parameters such as gain, shutter and signal-to-noise ratio are continuously optimized, and core test indicators (MTF, distortion, signal-to-noise ratio, light source brightness, uniformity, color temperature, comprehensive evaluation of system imaging capabilities, etc.) are used to comprehensively select light sources, lenses, and cameras, allowing many previously difficult imaging issues to be continuously broken through.

2. Image processing and pattern recognition develop rapidly

   In image processing, with the extraction of high-precision edge information from images, many low-contrast defects that were originally mixed in background noise and difficult to detect directly began to be resolved.

   In terms of pattern recognition, it can be regarded as a labeling process. On the basis of certain measurements or observations, the patterns to be recognized are divided into respective patterns. Decision theory and structural methods are mainly used in image recognition. The basis of the decision theory method is the decision function, which is used to classify and identify pattern vectors, which is based on timing description (such as statistical texture); the core of the structural method is to decompose the object into patterns or pattern primitives, and different object structures have different primitive strings (or strings). By using the given pattern primitives for the unknown object to find the encoding boundary, obtain the string, and then judge its category based on the string. In terms of feature generation, many new algorithms are constantly emerging, including features based on wavelets, wavelet packets, fractals, and unique component analysis; correlator support vector machines, deformed template matching, and the design of linear and nonlinear classifiers are constantly being extended.

3. Breakthroughs brought by deep learning

   Traditional machine learning mainly relies on humans to analyze and establish logic for feature extraction, while deep learning simulates brain work through multi-layer perceptrons and builds deep neural networks (such as convolutional neural networks, etc.) to learn simple features, create complex features, learn mapping and output. During the training process, all levels will be continuously optimized. In specific applications, such as automatic ROI area segmentation; punctuation positioning (unknown defects can be flexibly detected through anti-authentic vision); re-detection of defects that cannot be described or quantified from heavily noisy images, such as orange peel defects; distinguishing true and false defects in glass cover inspection, etc. As more and more machine vision software based on deep learning is introduced to the market (including vidi from Switzerland, SUALAB from South Korea, ASTRI from Hong Kong, etc.), the empowerment of machine vision by deep learning will become more and more obvious.

4. Development of 3D vision

   3D vision is still in its infancy, and many applications are using 3D surface reconstruction, including navigation, industrial inspection, reverse engineering, surveying, object recognition, measurement and grading, etc. However, accuracy issues limit the application of 3D vision in many scenarios. The first application in engineering at present is the volume measurement of standard parts in logistics. I believe this has huge potential in the future.

The above is the development trend of machine vision inspection technology introduced to you by the editor of Dior Technology. Machine vision has a wide range of industrial applications. Its core functions include: measurement, detection, positioning, identification, etc., technology pioneers and product developers in the field of photodynamics, medical lasers, medical machine vision systems and products, mainly including: a full range of industrial cameras, special cameras for industrial image measurement, special crosshair cameras for laser cutting, special image collectors for microscopic imaging, ultra-high-definition pathological microscopic imaging conference systems, pathological microscopic image analysis and remote consultation systems. Products are widely used in scientific research institutes, hospitals and schools, factories and mining enterprises, cultural relic appraisal units, criminal investigation and justice, intelligent transportation and other industries.