LEVEL CONTROL USING IMAGE RECOGNITION
bharadwaj874@gmail.com
Abstract--. The main aim of the work undertaken in this paper is to present a digital device for level control using image recognition. Integral to process control in many industries, different types of level measurement sensors used like ultrasonic, capacitance, hydrostatic pressure, electromechanical, gamma types. All these are of analog sensors rather than a digital. A simple webcam with roborealm software is used for sensing level and control. Matlab takes some time delay to track, recognize the live process & delay in generation of commands for control application so problem arises for live monitoring & control, whereas Roborealm software has an advantage that it recognizes & tracks live objects faster and generates commands for control applications. The position of Webcam is in horizontal for transparent tank or on top of tank for non-transparent. A microcontroller based controller circuit designed for tank level control. The use of object tracking with image recognition achieved greater accuracy & resolution is in the order of five times the capacitance type level sensor. The method developed shows the effectiveness of the scheme employed.
Keywords-- Roborealm, image recognition, RGB filter, COG module, calculate distance module.
BLOCK DIAGRAM:bharadwaj874@gmail.com
Abstract--. The main aim of the work undertaken in this paper is to present a digital device for level control using image recognition. Integral to process control in many industries, different types of level measurement sensors used like ultrasonic, capacitance, hydrostatic pressure, electromechanical, gamma types. All these are of analog sensors rather than a digital. A simple webcam with roborealm software is used for sensing level and control. Matlab takes some time delay to track, recognize the live process & delay in generation of commands for control application so problem arises for live monitoring & control, whereas Roborealm software has an advantage that it recognizes & tracks live objects faster and generates commands for control applications. The position of Webcam is in horizontal for transparent tank or on top of tank for non-transparent. A microcontroller based controller circuit designed for tank level control. The use of object tracking with image recognition achieved greater accuracy & resolution is in the order of five times the capacitance type level sensor. The method developed shows the effectiveness of the scheme employed.
Keywords-- Roborealm, image recognition, RGB filter, COG module, calculate distance module.
SCHEMATIC DIAGRAM
GRAPH:
ORIGINAL PICTURES:
LEVEL MONITORING SCREEN :
OPERATION:
The experimental setup consists of a webcam arrangement either on top of tank or in horizontal to tank that depends on transparency of tank. A green colored float is placed in the tank. RGB filter removes all surrounding colors except float. The convex hull module will transform a binary blob into its convex hull & it is used to simplify the shape of a binary blob by simplifying the perimeter of float, which reduces lighting effects of environment. The smooth hull module is used to be average blobs perimeter within the specified window. The final outline is then weighted against the new averaged outline and the original. Convex hull & smooth hull is used to reduce outside noise and achieved better tracking. The Mean filter is used to soften an image by averaging surrounding pixel values.
The experimental setup consists of a webcam arrangement either on top of tank or in horizontal to tank that depends on transparency of tank. A green colored float is placed in the tank. RGB filter removes all surrounding colors except float. The convex hull module will transform a binary blob into its convex hull & it is used to simplify the shape of a binary blob by simplifying the perimeter of float, which reduces lighting effects of environment. The smooth hull module is used to be average blobs perimeter within the specified window. The final outline is then weighted against the new averaged outline and the original. Convex hull & smooth hull is used to reduce outside noise and achieved better tracking. The Mean filter is used to soften an image by averaging surrounding pixel values.
Center of Gravity (COG) module locks the float and calculates the co-ordinates of present float position. The initial co-ordinates of float saved in a variable, when the tank is empty and save current co-ordinates in another variable. Now the calculate distance module provides a way to easily calculate the distance from the two variables. The X co-ordinate is making null value in calculate distance module. Now we get the current level of tank in terms of distance. A VB script program is used to convert distance to level of tank. The set point of tank level stored in VB script. A simple if & else statements are used for control application. If current level less than set point then send a command via serial port to 8051 microcontroller to open the control valve using a 12v dc motor controlled by L293D motor driver else close the control valve when current level exceeds set point.
CONCLUSION:
The use of image recognition for level measurement and control achieved greater accuracy & resolution is in the order of five times the ordinary analog capacitance type level sensor.
