A Method for Defect Detection of Objects Using Acoustical Signal Processing

Objects which are made of glass, porcelain, steal, plastic, ceramic, etc. play a vital role in industry and in households. These objects can be damaged during the production, store-in, and delivery. It is very important to deliver damage-free or non-defective objects and even the expired food or beverage packages can be considered as defective items. Therefore, in this research, a method is proposed to overcome costly systems available in industry for detection of defective objects through frequency spectrum analysis. An algorithm is developed to extract features to make the comparison between defective and non-defective items. The frequency spectrum is obtained using the Fast Fourier Transform (FFT) and the features are extracted using Cross-Correlation and Mel Frequency Cepstrum Coefficients (MFCC) to identify a defective item from a good one. The comparison process involves the use of the Euclidean distance which measures the percentage of dissimilar bits out of the number of comparisons made. The proof of the results of this investigation is given by testing samples of glass container. Therefore, this methodology can be applied for inspecting any kind of defective item as mentioned above.