Eggshell crack detection based on information fusion between computer vision and acoustic response

In order to increase the detection accuracy of eggshell crack, information fusion technology of computer vision and acoustic response was introduced for eggshell crack detection. An experimental system including computer vision system and acoustic response system was built up. Firstly, the acoustic response signals were captured and analyzed, then six variables including the dominant response frequency (f1, f2, f3, f4), the mean value of coefficient of skewness (CS), and the mean value of coefficient of excess (CE) were extracted after eggs was impacted four times on eggshell equator. With the six variables as inputs, 6-15-1 BP neural network was built to detect eggshell cracks. Secondly, the eggshell images were captured and processed through computer vision system, and five geometrical characteristic parameters of crack and noise regions of eggshell images were extracted. With the five variables including area (A), roundness (R), major axis (Max), minor axis (Min) and the quotient of long path and short path (LS) as inputs, 5-10-1 BP neural network was developed to detect cracks and classify eggs. Finally, the eggshell cracks were evaluated based on the difference of detecting results between computer vision technique and acoustic response technique. The results showed that the detection accuracy of cracked eggs were 92% and 68% respectively by computer vision technique and acoustic response technique. However, the accuracy equaled to 98% by the information infusion of two techniques. The information fusion technology was better than single technique, and the method based on the information fusion of computer vision and acoustic response was applicable for detecting egg cracks.