No abstract

In order to check and eliminate the cracked eggs quickly and exactly in the TMS320VC33 DSP as a hardware platform, and has many functions such as controlling knock-equipment, sampling and processing the acoustics signals, judging the cracked-shell eggs, communicating with the host computer etc. In order to improve the detecting accuracy, 6 characteristic parameters were extracted in cepstrum domain besides the common used characteristic parameters in frequency domain as candidate distinguishing factors. Step discriminant was used to optimize the combination of these characteristic then established the optimum discriminant functions of normal eggshells and cracked eggshells respectively according to the step discriminant result. The accurate recognition ratio for cracked eggs is up to 95.5 percent, and the system can check 12 eggs per sec, which satisfied the requirements of the process of egg products.