基于量子核聚类算法的图像边缘特征提取研究

为了提高图像边缘特征提取质量,采取了量子核聚类算法。首先把像素映射量子编码,在码元建立域内对像素块进行随机采样;然后通过聚类距离计算数据点和每一个聚类核心的距离,把数据向量分配到距离最小的核心向量中,核函数确定有效影响范围;最后对像素聚类相异性分析,给出了算法流程。实验仿真显示这种算法对图像边缘特征提取轮廓清晰,连贯性好,评价指标MS和聚类准确率较好,算法收敛快。

Image Edge Feature Extraction Research Based on Quantum Kernel Clustering Algorithm

To improve the quality of image edge feature extraction,quantum kernel clustering algorithm is proposed. Firstly, pixel mapping quantum was coded and pixel blocks in encode element domain were sampled randomly. Secondly, the distance between data point and every clustering core was calculated by clustering distance, data vector was distributed to the core vector at the minimum distance, and valid effect range was determined by kernel function. Finally, pixel cluster dissimilarity was analyzed and procedure was described. Experimental result shows that quantum kernel clustering algorithm can achieve the image edge feature extraction with well-defined profile and high consistency, good assessment factor MS and clustering precision, and fast convergence.