Machine Vision

/ Research

Machine Vision


Center for Nondestructive Evaluation

Machine Vision is the technology that uses images as its source of information and act as an intelligent identifier/locater in various Industrial and Robotics applications.

Works being carried out-
Camera based attendance:
This system is developed to overcome the drawbacks in the conventional attendance marking system. The system extracts the detected faces in an input image and compares with the images in the database. Based on the comparison the corresponding faces are recognized and the attendance is automatically marked to the respective faces, hence ensuring a foolproof, robust attendance marking system.

Miniaturized Industrial conveyor belt
A camera is mounted in the path of the conveyor belt such that the objects passing the belt lies within the field of view of the camera. The camera detects various objects and classifies the objects as wanted or unwanted objects. Based on the camera recognition of objects, an actuator in the rejection mechanism is activated to reject the unwanted objects.

Finger print identification
The system recognizes the biometric features and classifies the persons based on their finger print. This system locates the edges of the input image and compares with that of the database image to identify a person.

Length and angle detection in real time images
The system can detect the length of an objects in real world units, which replaces the use of conventional rulers. The system also identifies the angle of intersection of various objects in the image.

Selective localization of faces
The system searches for a specific person`s face in an image of group of persons and localizes the required person by marking his face.

Object recognition
This system uses a standard pattern recognition algorithm to identify the object of interest. The template of the required object is stored in the database and the system identifies if the object is found in the input image.

Automatic object counting
The camera detects various objects with unique colors and segments the objects based on its color threshold. After segmentation, the number of objects of each color is automatically counted.