DIP Based Beverage Inspection on Conveyor

This project aims at the development of a system for the removal of faulty product in beverage, water bottles industries. It can be self-operational in controlling, starting with bottle detection as a faulty product then pushing the products away from the conveyor to some other basket. Our project m

Project Title

DIP Based Beverage Inspection on Conveyor

Project Area of Specialization

Artificial Intelligence

Project Summary

This project aims at the development of a system for the removal of faulty product in beverage, water bottles industries. It can be self-operational in controlling, starting with bottle detection as a faulty product then pushing the products away from the conveyor to some other basket. Our project mainly deals to detect and remove the faulty products form the product line (conveyor).

This work is currently done manually in product lines in industries. This project is divided into two major portions i.e. detecting bottles on a conveyor using IR sensor and synchronize it with the movement of conveyor using the electrical circuit part that will include Raspberry Pi microcontrollers and once the IR detector detects bottle on conveyor it will give the signal to Raspberry Pi and start the camera

Another part is the camera for image capturing and of course Raspberry Pi for doing image processing with Open CV libraries. Raspberry Pi will be used for image processing and sending commands to the mechanical parts to work accordingly. The camera will take a real-time image of bottles and will send it to Raspberry Pi where it will be matched with predefined data for template matching.

This project will cover three major inspections task i.e. bottle cap detection that either it is properly locked or not, logo detection that either bottle has a logo or not and Third is level checking of liquid inside the bottle that either it is filled to a specific level or not.

Project Objectives

Project Objectives

The objectives of this project are:

• Hardware implementation of the conveyor

• Implementation of a motor driver circuit and IR sensors in the circuit

• Detect bottles on a conveyor using IR sensor

• Controlling of conveyor using microcontroller

• Detect whether loose or missing caps or not by using adaptive thresholding and contouring.

• Detect whether the bottle has a logo or missing logo by HSV colour detection and template matching

• Detect the level of liquid inside the bottle by image filtering.

Project Implementation Method

This project contains two parts hardware and software.

Hardware part

The Overall system is the integrated system which contains conveyor, motors, raspberry pi3, camera, pusher, motor drivers and IR sensors Raspberry Pi3 is controlling the whole device making logical decisions and implementations.

When the bottle is placed on the conveyor sensor senses it. Binary input is taken from the sensor and the controller turns the camera on. Then inputs from the camera are sent to Raspberry Pi and Algorithms of image processing are applied. Stepper motor attached to pusher will be inputted by the controller if a defected bottle is detected. If any of three faults are not detected, the bottle passes smoothly through the conveyor and this process continues till the conveyor remains on.

Drv8825 motor driver is used to control stepper motor while LD293 is used to control DC motor

Software Part

 In Software, Part Image processing algorithms are been applied in order to detect the faulty object moving onto the conveyor. For this, Python with OpenCV has been used. Ubuntu-mate Operating system is being used for raspberry Pi3. Another software XMing and IP scanner to connect raspberry pi to the computer. Decisions made by the controller are based on the results of image processing.

Image Processing Algorithms 

Image Processing Algorithms are used to detect the following three faults in the bottle

1. Faulty Cap
2. Missing Label
3. Liquid Level

For the cap, we have applied grayscale effects and then image segmentation using adaptive thresholding. Then binary inversion is done. After those morphological operations are applied. Area of connected components is calculated with the help of contour function to check the cap position.

For label detection, after defining the region of the label, HSV  colour segmentation is applied. Masked Image is then achieved and Morphological Operations are used to remove noise. If a specific colour of the label is present then a label is available. Another Technique Template Matching is also used for label detection.

For level of liquid detection, after defining region of level, Gaussian Blur and Median blur algorithms are applied for noise removal. Then Thresholding using Adaptive filtration is done. The masked image is achieved and Hough Transformation is applied to check line parameters.

Benefits of the Project

With the automatic accurate inspection system industries can improve their yield ratio as a greater number of products can be inspected in less time in contrast to accuracy, efficiency, cost-effective. As machine vision is used in such system which will work on specified data which will be given as a predefined data. So, accuracy can be improved. Moreover, industries can use such a system for maximum time as humans can only work to some extent. Moreover, this system can be extended to more inspection works like as we are working on a major three task. It can be used in different industries like in textile industries to inspect the textures on their products. Furthermore, in our domain, it can be used to check whether bottles have cracked, or they are crack less.

Technical Details of Final Deliverable

1. Any Source device must be connected to the socket.
2. IR sensors are used to detect the object, Binary input from the sensor is sent to Micro-controller
3. The connection between Micro-controller (Raspberry Pi 3) is developed with other components such as web-camera, pusher, motor drivers in order to control the system.
4. Micro-controller turns the camera based on the Binary output of the sensor.
5. Web-camera is fast enough to capture real-time images.
6. Inputs from the camera are stored in a separate database.
7. Image Processing Algorithms are be applied to the real-time image from the camera.
8. Image Processing Algorithms detects faulty cap, label and the liquid level of the bottle.
9. The microcontroller makes decisions on the bases of image processing and sends commands to the stepper motor 
10. The device sends feedback to the controller after removing the faulty bottle

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

Food , Manufacturing

Technologies

Artificial Intelligence(AI)

Sustainable Development Goals

Quality Education, Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com