Research On Nondestructive Detection Technology Of Walnut Quality

Traditional detection methods were unable to detect walnut quality completely. The theory and method for detection on internal and external quality of walnut using soft X ray technique and computer vision were discussed in this paper. Main contents of the study were as follows:1. The best parameters of soft X ray equipment for quality detection of walnut was determined through experiments, which were as follows, voltage was 65 kV, electric current was 0.75mA, and the integrated time was 2.67ms.2. Attenuation coefficients of each component of walnut with respect to soft X ray were studied. Results showed density determined the attenuation coefficients, The higher the density, the bigger the attenuation coefficients.3. It’s difficult to extract fine details in soft X ray image. A novel method named integrated method was developed to resolve this kind of problem. Laplace operator was first employed to enhance initial image, the result image was added with the original image. The details of the image was enhanced. Then Sobel operator was applied on the original image to enhance the edge. Subsequently, mean filter was used to remove the noise.Finally, adding the details enhanced image and the edge enhanced image to the original image to produce a new image which was smoothed by mean filtering.The new method developed was proven to work well in noise removal and details enhancement. The accuracy rate of intact and empty walnut was 91.8%.4. The visible light computer vision technique was studied for detection of broken walnut due to low detection accuracy rate based on soft X ray technique. At first the G(green) component of the image captured was extracted for segmentation of broken area using threshold method. The result showed the accuracy rate of broken walnut and intact walnut was 71% and 87.1% respectively.Soft X ray technique and computer vision technique were employed to detect walnut quality including internal and external quality in this study. It produces a new method to detect walnut quality fast, nondestructively and completely and provides a feasible approach for detection of similar varietites of nut.