Development of Object Recognition Technique using Quantum Information Processing

在本研究中，我们评估了量子物体识别算法对大量非结构化、非正交数据的物体，如数字图片和数字电影的识别效果。在所开发的量子物体识别算法中，利用量子纠缠和量子叠加来处理数据，处理速度非常快。为了将该算法应用于大量的目标，提出了一种基于加速稳健特征(SURF)和k-均值的自动特征提取方法。作为大量的对象，使用了诸如字母表之类的1001个图像。对每个原始图像提取四个特征图像。结果，检测到正确图像的平均概率幅度为0.92。实验结果表明，该算法对大量目标具有较高的识别能力。

In this research, we evaluated the effectiveness of the quantum object recognition algorithm to a large number of objects such as digital picture and digital movie that are unstructured and non-orthogonal data. In the developed quantum object recognition algorithm, quantum entanglement and quantum superposition are used to process the data with very high speed. To apply the developed algorithm to a large number of objects, a method of automatic feature extraction using SURF (Speeded Up Robust Features) and k-means is developed. As a large number of objects, 1,001 images such as alphabet are used. Four featured images to each original image are extracted. As a result, average probability amplitude of the detection of the correct image is 0.92. We confirmed that the developed algorithm has high recognition capability to a large number of objects.