machine vision
machine vision


An ability of a computer to see is called machine vision. Cameras play the role of eyes in the overall system. The number of cameras and their resolution depends upon the application. In addition to cameras, analog to digital conversion and further digital signal processing is the technique that qualifies a signal to feed in a computer or robot for real-time applications.

There are a plethora of applications of machine vision in industries and our daily lives like process control, inspection, and guidance of robots. These systems use complex algorithms to control the system efficiently. The main input of the system is images captured by the cameras just like human eyes. The computer itself behaves like the brain and does further processing to perform the task intelligently as it learned while development.

Because of the intelligent behavior of machine vision, it has a lot of importance in the automation of industry. Where it is being used for guidance the robots to make synchronization between different sections of production lines and for inspection of the final products to minimize the fault positive errors.

History of Machine Vision

The history of Machine Vision dates back to 1930 when the first time a real-world problem was solved by using the electronic sorting machine. This machine was used for sorting food in New Jersey. This machine was developed by using different kinds of photodetectors. Similarly, in 1801 Joseph Marie Jacquard introduced punched paper cards that allow the loom machine to weave cloths with different designs as per the punched pattern on cards. The figure below shows the card and corresponding waved cloths.

The development in the field of machine vision made at immense speed when cameras and other high-speed computers were introduced in 1980. However, in 1959 NCR 102D computer was used to develop digital imaging devices by NASA. The system filled a room. Since then the image processing became of the core interests of researchers. Several algorithms developed by the researcher in 1969 and 1970 are being currently used. In 1970 SIPI’s researchers developed basic algorithms for image processing and defined the method of feature extraction, image de-blurring and image codding, etc. The processing speed of the existing system was very challenging at that time. In the mid-1980s, many companies invented special image processing hardware. These machines considerably maximize the speed of the systems but at the same time, it was costly. With the advent of PCs, display controllers and frame grabbers, software with image processing libraries began emerging during 1980-1990.

Today different software companies have invented a different type of software and libraries to perform different types of tasks on digital images. For example, Stemmer Imaging (By Puchheim) software package has eased the processing of very sophisticated images. Eventually, these all supports have enabled scientists to develop artificially intelligent devices for automating industries.

How does a machine vision system work?

The overall system involves the following three steps.

Capturing Images:

For capturing the images the selection of cameras depends upon the application for which the system will be used. Similarly, the number of cameras also varies as per the requirements of the application. While the selection of cameras two factors plays an important role one is resolution and the other is sensitivity. These both are interdependent. Here sensitivity is the ability of a camera to see in low light and detection invisible wavelengths. On the other hand, the extent to which the camera can differentiate between objects is called the resolution of a camera. These two factors are inversely proportional to each other, by increasing the resolution the sensitivity decreases and by increasing the sensitivity the resolution decreases. The cameras are can detect a much wider bandwidth of wavelength as compared to the human eye (a human eye can detect from 390 to 770 nanometers). Now cameras are being used in machine vision for detecting the other wavelengths that include Ultraviolet and infrared bands. 

Processing on Images:

After capturing the images from a video camera. An analog to digital converter (ADC) converts the analog signal to a digital signal. Several digital signal processing techniques refine the signal and provide a refined signal to a computer for further actions. For performing these actions in run time a high definition computer, suitable Random Access Memory (RAM), and graphic cards are required. Moreover, for the calculation of depth, we need artificial intelligent programs.


These are those activities that can be performed on the application layer. Here the actions can be performed as per the application. For example, if we want to make an artificially intelligent robot by using machine vision then different controller or processor cards (Raspberry Pi, Jetson nano or Up board, etc.) can be used to interface with a machine which can control the different actions of a robot as per the programming.

D/f b/w machine vision and Computer vision

Machine Vision term is used for the application with industrial applications such as the automation in production line and inspection of the product while finishing in the factory. On the other hand, computer vision can be categorized as a technology in which a computer takes an image and converts it to a digital signal and performs an appropriate action as per the programming.

Examples of Machine Vision:

There are a plethora of today’s applications that use machine vision. Some of them are below:

Electronic Component Analysis

Different companies are using a machine vision technique for analyzing electronic components. Like providing the datasheet by just visualizing the number and shape of the electronic components of the components.

Signature Identification

There are many types of signs that have different shapes. Machine vision can detect the signature of a particular person or company by just scanning.

Object and Pattern Recognition

The object’s pattern recognition can be done by using the state of the art machine vision algorithms.

Similarly, there are other industrial applications such as material inspection and object packing that can be done very efficiently by using machine vision.


To recapitulate the whole discussion, it is note that machine vision and artificial intelligence are changing the industries at a very high pace. Moreover, not only industries are getting benefits but other fields of life are also getting change. It is changing the way of living of a common person and there is not any confusion that in near future the machine vision will help in boosting the productivity of industries to fulfill the requirement of the world.

You may also know: Artificial Intelligence in automated trading


Please enter your comment!
Please enter your name here

seven + eight =