Product quality is one of the critical issues to be competitive in the global market place. Especially, food production companies are strongly required to control high product quality in their production lines. The contaminations such as metallic, or plastic, organic materials are harmful to human body and should be eliminate as much as possible. Therefore, machine inspections, such as X-ray or fluorescence spectrum methods work effectively to eliminate these contamination substances. However, the final inspections are taken care of by human inspectors to make sure the product quality kept in the required specifications. Contaminations are not perfectly eliminated by machine inspections and human inspections could cover some of them. That is not the only reason why the human inspector takes care of. Even though there is no contamination in the products, appearance of products may decrease the quality of the products. For example, even though air bubble has nothing to do with the taste of Tofu, customers do not purchase such Tofu which contains many air bubbles, whereas they would buy Tofu with some air bubble. Therefore, human inspectors review the Tofu package to check the defective products because of the air bubble, which is hard to be processed by machine inspection. This study proposes image processing-based air bubble detection method on Tofu packages to inspect the product quality without image sensor devices. Based on the results of air bubble detection on the Tofu package, the evaluation will be made on each package. The study applied an evaluation criterion based on the experiments. However, the results are not always identical to those by human inspection because of the disagreement of threshold value in evaluation. This paper presents the image-processing air bubble detection method to determine the quality of Tofu products and discusses the feasibility of this method in comparison with the human inspection results.

This content is only available via PDF.
You do not currently have access to this content.