AN INFRARED AND LASER RANGE IMAGING SYSTEM FOR NON-INVASIVE ESTIMATION OF INTERNAL TEMPERATURES IN CHICKEN BREASTS DURING COOKING

Abstract

Foodborne diseases caused by undercooked poultry fillets are significant problems that have changed the direction of research toward the assessment of internal temperatures in poultry meat during cooking. In this research, a novel infrared and laser range imaging system is proposed to estimate the internal temperatures of chicken breasts during cooking. It consists of three subsystems: an infrared imaging system, a laser range imaging system, and a neural network modeling system. The infrared imaging is used to determine the surface temperature of chicken breasts during cooking. The laser range imaging is used to reconstruct the 3D images of chicken breasts. Then, the neural network model is developed to predict internal temperatures in chicken breasts based on surface temperatures and geometric information. The experimental results show that geometric variables play an important role in internal temperature estimation. By using three time-lagged sequential infrared images and the geometric information, the system can predict the temperature with the accuracy of 1.54°C for mean absolute error, 2% for mean absolute percent error, and 3.08(°C)2 for mean square error. The combined infrared and laser range imaging show the potential for real-time, non-contact, and non-invasive estimation of internal temperatures of chicken breasts. It has no cross-contamination and can be used as a new inspection tool for enhanced food quality and safety.