联合变换相关识别是研究图像识别常用的一种重要方法。当探测器对目标图像识别时,从不同的角度、不同的高度获取的目标图像通常存在角度旋转和尺度缩放的畸变情况,如何使相关识别系统具有畸变不变的功能,是一个重要的研究问题。本文基...联合变换相关识别是研究图像识别常用的一种重要方法。当探测器对目标图像识别时,从不同的角度、不同的高度获取的目标图像通常存在角度旋转和尺度缩放的畸变情况,如何使相关识别系统具有畸变不变的功能,是一个重要的研究问题。本文基于综合判别函数用联合变换相关器对角度旋转和尺度缩放两种畸变图像的相关识别进行了研究,给出综合判别函数模板的合成方法。相关识别结果表明,这种基于综合判别函数的联合变换相关图像识别方法对尺度缩放、角度旋转畸变图像具有相关不变性,能够很好地完成训练样本中各图像的畸变不变识别,对于一定小范围内的角度旋转和尺度缩放畸变图像识别率可达到100%,证明了这种识别方法的有效性和可行性。The recognition function of the joint transform correlation is the important and commonly used images recognition method. The object images are always rotation and scale distorted images, when the image detector takes object images at different angles and different heights. The important problem that how to realize the distortion invariant image recognition of the joint transform correlation is constantly studied by many scientific researchers. The correlation recognition based on synthetic discriminant function is described and applied to settle the rotation and scale distorted images correlation recognition and the synthesis pattern method of synthetic discriminant function is given in this paper. The correlation recognition experiment results of distorted images show that the correlation recognition method based on synthetic discriminant function can realize effectively the distortion invariant pattern recognition of rotation and scale distorted images that are in the training image samples, and the recognition rate of images that are rotation and scale distorted within a certain small range is almost 100%, the results show that the images recognition method is effective and feasible.展开更多
文摘联合变换相关识别是研究图像识别常用的一种重要方法。当探测器对目标图像识别时,从不同的角度、不同的高度获取的目标图像通常存在角度旋转和尺度缩放的畸变情况,如何使相关识别系统具有畸变不变的功能,是一个重要的研究问题。本文基于综合判别函数用联合变换相关器对角度旋转和尺度缩放两种畸变图像的相关识别进行了研究,给出综合判别函数模板的合成方法。相关识别结果表明,这种基于综合判别函数的联合变换相关图像识别方法对尺度缩放、角度旋转畸变图像具有相关不变性,能够很好地完成训练样本中各图像的畸变不变识别,对于一定小范围内的角度旋转和尺度缩放畸变图像识别率可达到100%,证明了这种识别方法的有效性和可行性。The recognition function of the joint transform correlation is the important and commonly used images recognition method. The object images are always rotation and scale distorted images, when the image detector takes object images at different angles and different heights. The important problem that how to realize the distortion invariant image recognition of the joint transform correlation is constantly studied by many scientific researchers. The correlation recognition based on synthetic discriminant function is described and applied to settle the rotation and scale distorted images correlation recognition and the synthesis pattern method of synthetic discriminant function is given in this paper. The correlation recognition experiment results of distorted images show that the correlation recognition method based on synthetic discriminant function can realize effectively the distortion invariant pattern recognition of rotation and scale distorted images that are in the training image samples, and the recognition rate of images that are rotation and scale distorted within a certain small range is almost 100%, the results show that the images recognition method is effective and feasible.
