An Object Recognition Method Based on Convolutional Neural Network and Naive Bayes

Abstract

The invention relates to a convolutional neural network and naive Bayes-based target identification method, which belongs to the field of image processing and mode identification, and is characterizedby comprising the following steps of: (1) determining coordinates (x ', y') of image pixel points after translation and normalization in a training set sample A to obtain a new training set A '; (2)training the convolutional neural network, and updating the connection weight i; (3) extracting a feature vector X; (4) taking the feature vector X as the input of a naive Bayes model, and establishing the naive Bayes model for training; and (5) taking the test set samples as input, and sending the input to a combined network for classification to obtain a classification result. According to the established target recognition method based on the convolutional neural network and the naive Bayes, the naive Bayes are used for replacing a Softmax classifier in a traditional convolutional neural network to achieve classification prediction, and feature information of feature vectors output by a convolutional neural network full connection layer is fully utilized. It can be known through multiple sets of data experiments that the method is accurate and reliable in calculation and relatively stable in data result, and a stable recognition method on the basis of guaranteeing accurate classification is provided for picture target recognition.

Description

technical field [0001] The invention relates to the field of picture processing and pattern recognition, and mainly relates to a method for recognizing picture objects. Background technique [0002] At present, for the problem of image target recognition, most technologies cannot achieve high accuracy, are unstable, and have poor robustness. Although some techniques can achieve high accuracy, they need to go through some complicated preprocessing steps. Taking handwritten digit recognition as an example, handwritten digit pictures often need to be distorted and deformed to expand the training set, and various jitter operations are simulated to preprocess the image. Although this can achieve a high accuracy rate, it will reduce the accuracy to a certain extent. Insufficient practicability, but can not meet the stability requirements. In classic pattern recognition, features are generally extracted in advance, and after extracting many features, correlation analysis must be ...

AI Technical Summary

Problems solved by technology

[0003] At present, digital recognition technology in image targets is often associated with economy and business, and the technology put into use must ensure a high accuracy rate, because if the recognition error, even a small mistake, may also trigger a series of business problems. Disputes, even bring huge losses, resulting in irreparable results

Images