基于阈值分割法增加识别点提高图像识别准确率的研究

Research on Increasing Identification Points to Improve the Accuracy of the Image Recognition Based on Threshold Segmentation

在图像分割算法中,灰度阈值分割算法是最常用的算法。因其算法复杂度小、实现条件单一,在图像识别技术中应用最广泛。灰度阈值分割算法适用于对目标主体和背景灰度级相差较明显的图片进行分割,当图片环境较为复杂、灰度级表现不明显时,其效果会依据环境而改变。因此,灰度阈值分割法对图像识别的准确率有很大的不确定性。在此基础上,提出一种通过增加识别点来提高图像准确率的方法。该方法采用在灰度阈值分割后,对灰度阈值图像进行边缘计算添加识别点标记图像特征的方式,对图像进行二次识别。通过反复试验并记录准确率,所得结果对比表明:增加识别点的二次识别方法相比传统灰度阈值分割法,准确率得到有效提高。

Among image segmentation algorithms,gray threshold segmentation algorithm is a classical algorithm. It is the most widely adopted in image recognition technology because of its simple implementation,low computational complexity and stable performance.Gray threshold segmentation algorithm is suitable for segmentation of images with obvious gray level differences between the object and the background.When the image environment is comparatively complex and the gray level performance is not obvious,its effect will change according to the environment and therefore there is a lot of uncertainty improving the accuracy of image recognition by gray threshold segmentation algorithm.On this basis,this research proposed a method of increasing identification points to improve the accuracy of image recognition.This method makes secondary identification by making edge calculation of gray level images and increasing identification points to mark image features after gray threshold segmentation. Through segmentation experiments of a large number of images and accuracy statistics,the comparison of the results shows that the accuracy of the secondary recognition method with increased ident.