Rice detection and classification method based on deep multi-view feature

Abstract

The invention relates to a rice detection and classification method based on deep multi-view feature. The method comprises the following steps: 1) division of a data set: dividing a rice drop image data set through a cross-validation method to obtain a training set and a test set; 2) construction of a network model: carrying out feature extraction on an input image through a deep learning method,and meanwhile, continually updating model parameters through stochastic gradient descent; and 3) model analysis and performance evaluation: inputting the test data into a trained rice classifier, andcomparing probability predicted value and classification threshold to obtain a rice detection and classification result. Compared with the prior art, the method is greatly improved in accuracy, and improves defects of rice feature extraction of a conventional method; and the method finishes calculation of husked rice yield of the rice according to the classification result.

Description

technical field [0001] The invention belongs to the technical field of rice detection and classification methods, in particular to a method based on deep multi-viewpoint feature detection and application of deep learning methods, and a multi-layer convolutional neural network. Background technique [0002] Rice is the grain crop with the largest output in my country. The sown area of ​​rice is about 30 million hectares, and the total output is about 180 million tons, accounting for more than 40% of the national grain output. The sown area of ​​rice in Yunnan Province is about 12,000 mu, and the output is 18,000 tons. Rice production accounts for 36% of the grain output. It is also a large province of rice sowing and production. Rice is an important basic and strategic material related to the national economy and people's livelihood. Therefore, in the purchase, storage, transportation of rice and the purchase and sale of policy-oriented grain, it is necessary to conduct on-si...

AI Technical Summary

Problems solved by technology

[0005] 2) The error of manual detection and determination is large

[0006] At present, the determination of the roughness rate needs the sense of the inspector to identify the imperfect grains. However, due to the different understanding of the standards, the subjectivity is strong, which leads to the increase of the error of the roughness rate.

[0007] 3) The efficiency of manual detection is poor

[0011] Looking at the research on rice grading in the past 10 years, most of the achievements revolve around how to improve traditional industrial control or use traditional image processing methods to grade and classify rice, but the results are not obvious

[0012] For the special problem of rice detection, relevant scholars have further proposed many novel and effective algorithms from the level of feature improvement. For example, the scale-invariant feature transformation algorithm is widely used. This algorithm determines the candidate matching by calculating the distance of the invariant feature vector. Point pairs, so as to match the graphics, but each feature point is represented by a 128-dimensional vector, and the amount of data to be processed is large, so there will be problems such as inaccurate control, slow calculation speed, and low precision of registration points.

However, these classical methods are prone to the situation that the number of feature points is small and the matching fails

Images