Research About Potato External Damage Identification Based On The Hyperspectral Image Technology

The hyperspectral image system is the preferred technology to detect the comprehensive quality of fruits and vegetables with characteristics of continuous multi-band, high spectral resolution and so on. Potato has the priority to the development because it is one of the strategic leading industry of ningxia. Potato’s quality detection is an important link of potato processing. Based on the hyperspectral image system, this paper researched how to identify potato’s external frostbite, mechanical damage, hurt and normal potato. The main research contents and conclusions of this paper are as follows:(1)We have done all band (468nm-1000nm) of the principal component analysis with original hyperspectral image data cubes of potato and chosen the principal component images with obvious features for the interested area. To shorten the processing time when the on-line identification, we have analyzed the contribution rate of each band of the chosen principal component images and selected the contribution rate of the top five wavelengths (480nm,676nm,750nm,800nm,960nm) bands as feature bands.(2)In order to adapt to the Bayesian classifier model’s request of approximation of Gaussian distribution for input data, we have smoothed the feature images with recursive least squares method.(3)We have built the Bayesian classifier model to identify potato’s external frostbite, mechanical damage, hurt and normal potato. We have established a total of4kinds of process (including whether having interested areas and whether smoothing images) for model to identify. And we compared them, considering model recognition accuracy and processing time, We selected non interest areas and not smooth process for identification. Except for the low identification in the potato’s class of frostbite (50%recognition rate in prediction set), model’s recognition rate is80%for normal potatoes,75%for mechanical damage potetees,90%for hurt potatoes in the forecast set. And this process realized the automation of the whole system.(4)This paper provides technical reference on hierarchical classification in front of the potato processing for potato processing industry.