首页>
外国专利>
Feature quantity extracting method, the object classification method, the object identification method, the feature amount extraction device, object classification system, a subject judging unit, a feature extraction and subject classification and subject determination program and a recording medium recording the program
Feature quantity extracting method, the object classification method, the object identification method, the feature amount extraction device, object classification system, a subject judging unit, a feature extraction and subject classification and subject determination program and a recording medium recording the program
From an image obtained by imaging an object, feature extraction methods representing a feature of the subject regardless of the shape of the subject, and provides a classification and discrimination method of an object using the extracted feature quantity. An extracting feature quantity extracting process characteristic quantity from the image of the subject captured by the imaging means, the step of creating a filter bank from the image, a step of creating a maximum brightness image from the filter bank, the maximum brightness Set the image area circle image, and the step of setting the center Cc and the radius Rc of the image area circle, the pixel having the maximum luminance image, between the center Cc (a) to the position of the pixel (x, y) and the distance L D, the ratio of the radius Rc, (b) the luminance value F D of the pixel (x, y), (c) the sum of the difference of the luminance values of the neighboring pixels, and is projected to the three-dimensional space in the axial a step of, by including the step of creating a frequency histogram from a pixel projected on the three-dimensional space, and extracts a feature quantity from an image obtained by imaging a subject.
展开▼
机译:从通过对物体成像而获得的图像中,特征提取方法表示物体的特征,而与物体的形状无关,并使用所提取的特征量提供物体的分类和判别方法。从成像装置捕获的被摄对象的图像中提取特征量,提取处理特征量,从图像中创建滤波器组的步骤,从滤波器组中创建最大亮度的图像的步骤,最大亮度设置图像区域圆图像,以及将具有最大亮度图像的图像区域圆的中心Cc和半径Rc设置在中心Cc(a)到像素(x,y)与像素之间的位置之间的步骤距离L D, Sub>与半径Rc之比,(b)像素(x, Sub> y)的亮度值F D Sub> Sub>(c) Sub>相邻像素的亮度值之差的总和,并通过包括创建频率直方图的步骤在轴向上投影到三维空间从投影在三维空间上的像素中提取像素,并从通过对被摄体成像而获得的图像中提取特征量。
展开▼