Digital Image Processing Technology Applied Research, The Precision Measurement Of The Geometric Quantities

The digital image processing technology is a behavior that satisfies person's visual mentality or application requirement through dealing with image information. Correctly disposing and measuring images that reflect measured parameters is the primary mission in geometric dimensional measurement utilizing digital image processing technology. Precision is the important content in research of image processing and measurement. In the image measurement domain, accurate recognition to edge of images is the prerequisite to pledge the measurement precision. And how to detect the image edges effectively under strong background noise, and then carrying on the split-hair accuracy measurement should be the most concerned topic in precision measurement domains of geometric parameters.In allusion to edge detection of machining hole (including shallow blind-hole), the paper firstly explains definition and characteristic of edge, then introduces and analyzes application of classical edge detection algorithms as well as the mathematical morphology in detecting edges of image detailedly and systematically. According to experiments, every sampling image exists shortcomings in different level such as low contrast of edges, low signal-to-noise ratio and complicated marginal type because of some physical factors such as collapse on the sides as well as the burr engendered in machining course and bottom glisten of the shallow blind-hole, which results that can't detect edges of blind-hole accurately with the classical edge detection algorithms. How to detect true edge of the hole in strong background noise is the key and difficult problem in the field of image processing.In order to solve this problem, the paper studies from two aspects on software and the hardware respectively. Firstly we researches the application of de-noise and edge detection with wavelet transform according to its excellent characteristic of multi-resolution analyses in aspect of detecting signal on the side of software. The algorithm not only comparatively finely reduces noise that arises out of reflection of the bottom of blind-hole, but also protects edges of the hole availably from results of experiments. And then on the other side of hardware, we meliorate the camera lens by decreasing its depth of field. We can draw a conclusion from results of experiments that the measure we take enhances contrast of edges of the blind-hole and improves the measurement precision of its consumedly.In order to achieve higher precision of position of the blind-hole, the paper deduces principle that fractionates pixel in EasyGauge algorithm. On top of it, the paper discusses influences to the measurement result because of locomotion and noise. We can educe that the algorithm could achieve stable and high precision measurement results under strongbackground noise from experiments.At last of the paper, we chiefly introduce the hardware equipment used to sample images and the software system that aggregates functions of image sampling, image processing and image measurement, etc. Finally we measured the through-hole, original blind-hole, blind-hole after wavelet transforming and blind-hole sampled after meliorating camera lens, and then analyzed and compared the measurement results.