基于增量符号相关的缺陷检测方法

传统的规格化互相关算法在计算机视觉领域中用得较多,但其运算速度不能满足实时检测要求.用基于增量符号相关的算法对两幅图像间的缺陷进行检测可有效地缓解规格化互相关算法计算量大的问题.通过图像像素值间的大小比较得到增量图像后,计算出增量符号相关值,将它显示为与相关值成正比的亮度值就得到了检测结果图像.实验证明该方法在图像有亮度或对比度变化时仍能正确地检测出缺陷位置.

Defect detection method based on increment sign correlation

Conventional normalized cross correlation(NCC) has been used extensively for many computer vision applications,but its operation speed does not meet on-line detection requirements for time-critical applications.A method using increment sign correlation(ISC) for defect detection between two images can decreases largely computational complexity of normalized cross correlation algorithm.Firstly,the increment image is computed by comparing pixel values,then increment sign correlation values are computed,last the result image is displayed as a function of the intensity whose value is proportional to the increment sign correlation value.The experimental re-sults demonstrate that the proposed method can detect successfully defect location in the case of intensity or contrast variations.