Quality inspection for cheese packaging using machine vision and image processing

Zhe Cheng, Purdue University

Abstract

The acumen and sophistication of consumers have created the increasing expectation for improved quality in food product, which is considered as the essential element of daily life. In turn, this has encouraged food producers to improve their quality monitoring by deploying enhanced computer quality inspection technologies [1]. The aim of this thesis is to design an efficient and adaptable algorithm to accurately and efficiently monitor the quality of the packaged cheeses on the assembly line. Computer vision and image processing methods were used to distinguish unqualified cheeses from a large amount of samples. The criteria for classification were consisted of two main aspects, similarity of cheese shape and leakage condition. Gray and binary cheese images were converted from the original pictures, which were captured by cameras. The cheese part was extracted from the background for shape analysis to generate its signature, which was then compared with the signature of the standard cheese shape to measure the similarity by the cross-correlation method. Cheese leakage in the remaining part was discovered by setting a certain range of RGB value, which was subject to the condition of light sources. Two thresholds were set to control the detection result, which was intended x to best match human perception. Sensitivity, specificity, accuracy and receiver operating characteristic (ROC) were used to evaluate the algorithm's performance.

Degree

M.S.E.

Advisors

Chen, Purdue University.

Subject Area

Food Science|Computer Engineering|Packaging

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS