Machine vision inspection is a rapidly developing branch of artificial intelligence (AI). According to the definitions of machine vision by the Machine Vision Division of the Society of Manufacturing Engineers (SME) and the Automation Vision Division of the Robotics Industry Association (RIA), machine vision is a device that automatically receives and processes an image of a real object through optical devices and non-contact sensors to obtain the required information or to control robot motion.
Simply speaking, machine vision is using machines instead of human eyes. Machine vision simulates the eyes for image acquisition, extracts information through image recognition and processing, and finally completes the operation through the execution device.
Traditional machine vision inspection technology requires representing data as a set of features or inputting them into a prediction model to obtain prediction results. This requires completing specific actions, making it difficult to adapt to future flexible production needs, especially in scenarios where defect types are complex, subtle, and background noise is becoming increasingly difficult to apply.
After being equipped with AI deep learning function, machine vision converts the original data features into a higher-level and more abstract feature representation through multi-step feature transformation, and further inputs it into the prediction function to obtain the final result.
Machine vision based on deep learning can combine the efficiency of machine vision with the flexibility of human vision in an ideal state, thus completing detection in increasingly complex environments, especially when involving deviations or extreme environments, meeting the stringent requirements of downstream for defect accuracy and universality.
