高效能的织物疵点识别算法及其实现

Realization of High-efficient Identification Algorithm of Fabric Defects

通过对织物结构的深入剖析 ,提出运用布纹的直线特征纹理 ,由概率统计生成直方图 ,有效地提取织物的特征波形 ,经大量试验检测证明特征波可靠稳定。基于波形参数对比能准确定位突变或异常位置 ,正确识别织物疵点。为简化识别算法可引入阀值过滤 ,允许正常波形参数通过而记录异常状态变化 ,用过滤单一的正常纹理来鉴别类型复杂的疵点 。

Through closely analysis fabric structure, application straight line characteristic grain of cloth is brought forward, histogram is generated from probability statistics, which can efficiently abstract fabric characteristic waves, many large of tests certify that characteristic waves are reliability and stability. Based on waves parameter, mutation or abnormal position can be located precisely and recognized fabric defects correctly. In order to simplification identification algorithm, threshold filtration is quoted, which permit normal waves parameter to pass, but record abnormal condition changes. Using single normal grain filtration discriminate complex fabric defects, which is a simple and high-efficient algorithm of fabric visual detection.