Simple structure, orderliness and few gray level of SMD (surface mounted device) resistor image are the prominent feature of SMD resistor units array, through binary resistor image, edge detection, line detection, with the correlation coefficient as a criterion for defects. A method for resistor flaw detection was proposed based on subgraph projection matching. The feature of resistor flaw was extracted on basis of the method of PCA (principal component analysis). Then the resistor flaw would be classified by SVM (support vector machine). At last, an experimental platform was built and the result verifies that the detection rate employed the proposed method is 92.5%, and the method meets the requirements on high accuracy and speed.