Research And Application Of Image Mosaic Algorithm Based On Compressed Sensing In Parts Surface Defect Detection

In view of diversity of parts surface defects, strong interference on the spot and low efficiency and unstable quality testing caused by the widely-used way of off-line artificial sampling for the parts detection in companies, the paper explores the Compressed Sensing(CS) description of defect images on the surface of parts and a fast matching and stitching algorithm by using machine vision and CS in order to improve mosaic speed and quality of multiple images on outside cylindrical surface of parts.In recent years, Image mosaic technique based on CS and Scale Invariant Feature Transform(SIFT) has gradually become the research focus in the field of picture processing.However, it is not ideal in some aspects like improving image matching accuracy, reducing the cost of time and space of image mosaic to ensure real time of display and meeting the authenticity of visual perception of human eyes. In order to solve the above problems, the research adopts a fast SIFT image mosaic algorithm by introducing CS technique, which gets a very good visual effect that it not only makes the matching of feature points more accurate,but also makes the mosaic speed faster.The author conducts the following researches:(1) The paper first makes a detailed review of image matching and image mosaic. The common image matching algorithms are concluded among which several typical image matching algorithms are introduced in detail. After that, several important transformation models in image mosaic are illustrated.(2) Then, Compressed Sensing theory is elaborated. Some basic contents which includes sparse representation of signals, design of observation matrix and restructuring algorithm of signals are presented and then their applications are analyzed and illustrated.(3) Considering the disadvantages of massive calculation and slow speed of traditional Scale Invariant Feature Transform(SIFT), an improved image mosaic method which combines SIFT and CS is proposed. This algorithm speeds up the feature points extraction and effectively solves the problems such as mismatch and crack in the process of image matching and stitching. The experimental result shows that both number of feature point and matching point in the proposed algorithm are less than the traditional SIFT algorithm. What’s more, the proposed image also takes less time.(4) According to the requirements of subject research project, an online hardware platform for parts defect detection is established and a image mosaic module for this testing system is designed based on Visual Studio platform. The improved algorithm realizes the detailed description of each step of image mosaic, which provides technical support for thefollow-up online defect detection of parts.