To enhance quality control process, we here proposed a system to detect defect in ceramic tiles. Accuracy to detect defect manually is lower due to human errors and industrial environment. We proposed an automated technique and methods to detect damaged tiles. As images are captured under different lightening conditions. We used a technique to remove light from image which helps to detect crack with good accuracy. Image pre-processing steps are used to remove noise from image. Convert the image to binary image. Read the binary image and create a vertical line shaped structuring element and Dilate the image with a vertical line structuring element. Dilation methodology will make the crack bolder. Due to image processing, system finds unwanted objects along with crack. To detect only crack we remove those unwanted objects from image using MATLAB inbuild function. Some gaps to be filled after removing unwanted objects. System detects crack based on connected component. System display text whether tile is damage or not. And display image with bounding box around the crack. The processing time for one tile was approximately 2 seconds. This outstanding achievement of results reflects that this automated system can effectively replace manual ceramic tile detection system with better accuracy and efficiency. The proposed system is able to detect damaged tile with 85% success rate.
- Reduce labor work
- Saves Time
- Provides high accuracy and efficiency
- Works only on good quality images