基于PCA和不变矩的三维物体识别算法研究

Research on 3D Object Recognition Algorithm Based on PCA and Moment Invariant

在计算机视觉问题的研究中,针对基于矩的目标识别算法实时性不高的问题,本文分析主分量分析法(PCA)在目标快速识别算法中的应用,提出基于Jan Flusser仿射不变矩和PCA融合的快速识别方法,即利用PCA在Jan Flusser仿射不变矩的特征空间中进行优化降维,减少了计算量,然后利用一些主流的识别算法对该方法的实时性和准确性进行验证研究。在MatLab平台下的仿真结果表明:本文方法的实时性在欧式距离上提高了23.68%,在概率神经网络上提高了8.7%,在支持向量机上提高了21.01%,准确性只有少量的降低,不改变识别算法的过程,且适合三维物体小角度变化下的识别。

For the problem of real-time of target recognition algorithm based on moment is not high,and analyzes the application of the principal component analysis（PCA） in fast target recognition algorithm,this paper proposes the method based on the Jan Flusser affine invariant moments and PCA of fast recognition.By using the PCA in the Jan Flusser affine invariant moment feature space are optimized for dimensionality reduction,the method reduces the amount of calculation,and then some mainstream recognition algorithms are used to verify the real-time and accuracy of the method.The experimental results show that the method of real-time in Euclidean distance is increased 23.68%,the probabilistic neural networks is improved by 8.7%,the support vector machine is improved the accuracy of 21.01%,and only a small amount of reduction in accuracy,it does not change the recognition algorithm of the process.Meanwhile,it suits for three-dimensional object recognition under small angle change.