| Potato is a major cash crop in northern China and an important cash crop,.The potato external quality nondestructive testing is an important part of the potato production and processing, but also for industrial production as well as the first step in potato processingthe tne paper through the use of potato hyperspectral images for non-destructive testing of potato tuber shape and the weight of a study and the main research contents are as follows:(1)Firstly, according to the national standards that potatoes are divided into deformity and excellent shape, then i use the hyperspectral imaging technology collecting acquisition the potato hyperspectral image information, and employ a variety of methods the potato grayscale images, denoising, filtering, preprocessing, according to the processing the results of selecting to choose the best results adaptive median filtering method for image preprocessing.(2) The preprocessed potatoes is obtained through the threshold segmentation,and it contrasted the resulet of the whole,part and iteratine global threshold segmentation.Then selecting the best mothed of iterative global thresholding,.By comparing various methods of edge detection, edge detection result of the comparison, the final choice to choose a batter edge detection based on wavelet.(3) After extracting the edge detection based on the wavelet transforms of the krawtchouk unchanged form of gragscale image. Using krawtchouk relatively unchanged form the turn away from the Euclidean distance and combine taxonomy classifies potato.(4) Formed the model of excellent shape for potatoes that based on the weight of the potatoes. By constructing potato gray image - gradient co-occurrence matrix, and then realized the optimal threshold grayscale image segmentation based on the maximum entropy principle, the statistics showed that the number of pixels potato, using polynomial fitting method to establish the potato area and weight relationship,AndPrediction error is less than the weight of 5g. |
PDF Full Text Request