Computer vision is an amalgamation of computer science, optics, mechanical engineering and industrial automation aimed at the mechanized extrapolation of the information present in visual images. Computer vision is an increasingly popular tool in many industrial contexts.
Related Categories
This very broad field focuses on the processing of imagery while practical applications, such as robotics and automation go further to connect the developed data with subsequent action. Though often used interchangeably with machine vision and robotic vision, these are not synonyms. Instead these fields embody two of the most prevalent applications of this technology within the industrial and manufacturing sectors. From informed symptom recognition in x-rays to controlled automotive welding, computer vision finds practical applications in a number of industries such as pharmaceutical, food processing, electronics, textile, medical, automotive and automation industries among others. In each of these applications the machine or robotic apparatus may be employed to perform specific tasks, but computer vision is designed with only two central objectives. The first is to create a three-dimensional structure from two-dimensional images. From this, the computer is programmed to recognize the contents of a scene and collect data on features, materials, size, etc. of a particular part or product. Beyond these two main purposes, the data collected can be used to count, sort, or inspect an object.
The many industrial uses for computer vision combined with commercial and even residential applications necessitate a wide variety of products designed for the recognition and evaluation of visual images. Despite the inherent variety, there are several components that are essential to the functionality of all vision systems. Vision sensors are used to detect when a scene is set up. In inspection applications, for example, this would be when the part or product to be analyzed is in proper position. The sensor triggers a smart camera or other image capturing device also connected to a frame grabber which converts the camera output and enters the data into the memory of the computer system. Vision software is then used to process the image and develop an understanding of the visual representations presented. To do this, the software removes static and optimizes the imagery before applying pre-programmed mathematical algorithms. These algorithms cannot match the comprehension and tolerance for variability found in human vision, but instead mimic this effect by comparing the object to a set of criteria in order to determine its purpose, value or level of imperfection. Highly advanced systems use recognition models which require extensive amounts of pre-programmed knowledge, logic and capacity for information storage and retrieval.
