Research On Car Wheel Rim Edge Recognition

With the development of modern industry, it has become an inevitable trend that the intelligent system to replace the traditional manual operation and it is an important research direction that the robot intelligent robot with image processing. As car rims polished rims robot needs to identify the car graphics, this paper focus on edge recognition technology for pattern recognition problems. Due to the two-dimensional image is the basis of image recognition, paper focuses on a two-dimensional image in image processing and feature extraction for image processing of two-dimensional images. The Edge reflected the image’s fundamental features which are simplest and most core basic features. Edge Detection in image recognition are discussed in detail in this paper.In order to resolve noise, the image edge missing and false edge interference problem, the paper present that classical edge detection combine with mathematics morphology and histogram equalization. It analyses the histogram equalization enhancement and morphology filter, combined with the identification of car rims in industrial processes. This thesis proposed a new image evaluation criteria noise parameter P and four combinations edge detector which combined traditional edge detection with histogram equalization. Finally, a new combinations edge detection is proposed by using noise parameter P to select and fuse the two algorithms better in the four kinds of algorithms.(1)The paper further analyze histogram edge detection and its improved algorithms, which is based on mathematical model theory of edge detection. It is found that the marginal integrity based on histogram of edge detection has been improved greatly.(2)Four combinations of edge detection algorithm is proposed, Basing on histogram equalization algorithm and mathematical morphology. Through different morphological algorithm for edge detection effect in four types of combined analysis, study Finds integrity and noise immunity performance of combination algorithm greatly improved.(3)On the basis of PNSR traditional image evaluation system, this thesis proposes a new image evaluation criteria which is noise parameter P. The Evaluation Criterion of strength lies in two points : First,The evaluation criteria can directly evaluate noise performance of the algorithm is good or bad. Second, evaluation criteria still can be used for images evaluation even if the image has been greatly transformed.(4)Combinations edge detection combine with noise parameter p for further improvement. This algorithm fuse combination algorithms for image enhancement and morphology together by using Anti-noise parameters P. Experimental results show that the image effect almost is the same whether or not have noise, and the integrity of edge is effectively improved. This combination methods own good balance in anti-noise performance, edge missing and false edge interference problem. This research provides some ideas for industrial processing and pattern recognition.