晚期肺结节图像特征准确识别仿真研究

Classification Techniques Application in Image Feature Recognition for Late Lung Nodules

在对晚期肺结节图像特征进行识别的过程中,由于病变区域的颜色和形状特征具有一定的模糊性,使得病变区域的不规则度受到模糊性干扰。传统识别模型,只分析肺结节的类圆形特征,忽略了这种干扰对结节图像造成的影响,一旦出现干扰,晚期肺结节图像特征无法实现有效识别的问题,提出采用分类识别技术的晚期肺结节图像特征识别方法,通过近似体积特征对圆形肿块特征进行描述,依据紧凑度与边界不规则度对棘状突起特征进行判断,通过统计矩向量和不变矩向量对结节特征和空泡特征进行描述。通过Mahalanobis距离对肺结节图像特征进行识别,判断。对经过识别确定为肺结节的图像区域进行求解,有效地完成了晚期肺结节图像特征的识别。仿真验结果表明,所提方法具有很高的检测精度,明显优于传统方法。

An image feature recognition method for late pulmonary nodules is presented based on classification technique. The features of circular lump is described through the approximate volume, the characteristics of spines are judged based on the compact degree and boundary irregular degree, and the characteristics of nodules and cavita- tion are described through the statistical moment vector and invariant moment vector. The image characteristics of pulmonary nodules are identified and judged By Mahalanobis distance, and recognition of image features for late lung nodule is effectively completed based on the solution for the image area of pulmonary nodules. The simulation results show that the proposed method has high accuracy, which is superior to traditional methods.