Error Detection in Rubber Bulbs

Introduction: Error Detection in Rubber Bulbs

About: Catch up with this tech geek on LinkedIn !

Industrial visits offer two things in common - Enjoyment & Exploration !

Well, on one such event, it was a visit to Jain Rubbers Private Limited which is situated about 50km from Chennai in the industrial estate of Gummidipoondi in Tamil Nadu that mooted an idea for developing a project. It produces Rubber Bulbs/Injection Sites whose applications are as follows :

  • Infusion and transfusions sets
  • Pharmaceutical Rubber Closures
  • Rubber Gaskets for Disposable Syringes
  • Rubber Plugs
  • Rubber Sleeves/Covers for Blood Collection Needles
  • Rubber Float Discs
  • Rubber Plugs for Y-Connection
  • Products for the Medical & Pharmaceutical Industry

Since, these rubber bulbs are subjected to manual analysis in the detection of errors, the process is indeed time consuming and tedious. There arises the need for Automation and the best possible solution can nothing be better than Image Processing using MATLAB.

Step 1: Analysis of Errors

The most common errors in rubber bulbs are as follows :

  • Short Bulbs
  • Black spots

The image shown corresponds to a perfect rubber bulb with no errors.

Step 2: Identification of Errors

The parameters of a proper rubber bulb -

  • Outer Diameter : 12.00 mm
  • Stem Diameter : 6.00 mm
  • Inner Diameter : 3.10 mm
  • Length : 47.00 mm
  • Material : Natural Rubber (JLB1) / Polyisoprene (Latex Free) (JIB1)

The attached images illustrate the common errors of short bulbs and black spots in the production of rubber bulbs.

Step 3: Running MATLAB Scripts

The different stages in the image processing of rubber bulbs in MATLAB are as follows -

  • Original image
  • Dilated image
  • Binary images in two sections
  • Warning note

Refer the images for better understanding !

Step 4: Insights on Project

The video link of the MATLAB automated script for image processing of rubber bulb is as follows :

Proposed Hardware Design

  • The rubber pieces are let in on a narrow lane into a string less than 2% of diameter of rubber to hold the rubber firm but not tight.
  • Rotation using a string like material is essential in this regard because a 360 degree view of rubber piece is required with a conveyer belt running all the time.
  • Rubber pieces are rotated through an angle of 120 degree by the string.
  • Cameras of 4-7 MP are required. • Three images are generated under a common file name for one single image.
  • The images are analysed using Matlab scripts.
  • Now the sweeper brushes push the inappropriate ones into the corresponding boxes.
  • Only constrain is the holding and leaving process of the string which can soon be rectified.

Conveyer belt

  • Two conveyer belts intact can be used as a profitable measure for 5 tracks.
  • Else, only a single conveyer belt on the whole can be used for a single track.
  • Black background is preferred.

Time prediction

  • For a sequence of 5 rubber pieces with 5 cameras, a total of 3 seconds for capturing the images, 3 seconds for rotation of string, 2 seconds for holding and leaving the string, 1 second as time lapse accounts to about 10 seconds to check for the identification of 5 pieces.
  • For 5 tracks and 25 cameras, 10 seconds for identifying the defects of 25 pieces which puts together 150 rubber pieces in a minute of time.

Need more insights on Image processing in MATLAB ? Just chase the link below -,d.dGo

Stay around for other projects ... Thanks !

Be the First to Share


    • Plastic Contest

      Plastic Contest
    • The 1000th Contest

      The 1000th Contest
    • Battery Powered Contest

      Battery Powered Contest