一种图像敏感对象的识别方法

A Recognition Method of Image Sensitive Objects

摘要:

针对现有的图像识别方法识别某一指定对象,提出了一种图像敏感对象的识别方法,该方法改进了现有目标识别的局限性;首先通过计算得到图像中对象的敏感度值,然后对图像对象进行纹理、色彩、形状和显著特征提取,最后结合敏感度值识别出图像中敏感度大的对象,并与经典的敏感度检测方法比较;实验结果表明该方法敏感信息丢失少,且能精确得到图像对象敏感度值,通过敏感度的不同,识别出人眼感兴趣的对象信息。

Abstract:

For the conventional image recognition methods, this paper proposes a recognition method of image sensitive objects which improves the limitations of present approaches. This method gives each object in the image a precise sensitivity value. Then it extracts the texture, color, shape and salient features of image object. Finally, according to the sensitivity values, it identifies the objects with large sensitivity value in the image and compares with the classical methods of sensitivity detection. Experimental results show the proposed method loses less sensitive information and can obtain more accurate image object sensitivity value and identify the objects of interest of the human eyes by different sensitivity values.