Study Of Computer Vision Applications In Industrial Inspection

The computer vision automatic detection system applies widely in the industry. It can replace human to complete a series of highly duplication and risk works. The system needn't to contact the objects, has quick processing speed, highly precision, can easy be expansion, and is very reliability. To some extended, it enhances the productivity and realizes the producing automation. The topic of the article is originates in "the automatic counting of industry components project" of some ordnance machine factory. It requires to study and make a computer automatic counting system to replace the manual work.The system uses common CCD camera and PC as its main hardware parts. The software is composed by the operating system and the specialized application software.It is the most important to develop the special application software, which can execute the function of automatic recognition and counting. And the difficulty is to separate the conglutination components.The new method of illumination normalization is region-based. It simply partition the picture into four equal regions, and according to the characteristic of the picture, combines the Histogram Equalization and the Gamma Intensity Correction methods for relighting. This method works out the unequal illumination problem, and enhances the precision in next segmentation step.An Image Segmentation Algorithm Base on Mathematical Morphology and Layered Features Extraction is bring forward in the image segmentation process. The algorithm combines the Layered segmentation and the Mathematical Morphology, realizes the elaborate segmentation of complicated image. The thought of Layered segmentation partition the different objects with same characteristics, and for the complicated objects, it means segmentation step by step.Do Partition with different objects individually. At first, with the high brightness of the parts' inverse, uses Histogram Threshold method to realize segmentation. Then extracts the obverse parts from the image and separate the conglutination components. When disposing the obverse parts, leaches the single ones and extracts the gray morphological gradient as the boundary of the parts, and separates them individually.After the accurate segmentation, use the connect area mark method to realize the automatic counting.