This paper presents a new non-contact, non-destructive method for automatic inspection of micro-structures. A piezoelectric actuator is used to vibrate the devices under inspection. The frequency of the vibrations is swept within an interval that includes the resonance frequency of the parts. A microscope camera is used to monitor the vibrations of the structures. At resonance, functional devices undergo large oscillations and these are seen as a blur on the camera. Two image processing algorithms are used to measure the degree of blurring in the camera images, namely, Quadtree segmentation and Otsu's adaptive thresholding method. Using the Quadtree segmentation method, blurring is detected from a drop in the count of image partitions. Using Otsu's adaptive thresholding method, blurring is detected from an increase in the count of white pixels in the binarized image. In both cases, by setting a threshold on the number of partitions (white pixels), resonant parts can be distinguished from broken ones. Experimental tests proved the effectiveness of the proposed method, and indicate the feasibility of its application in a real time plant environment. This is believed to be the first fully automated quality control method for micro-structures based on the analysis of their resonance vibrations.