Feature Extraction And Analysis For Envelope Image

As the development of computer science, image processing technology plays an important role in our regular life. The image, processed using an algorithm, brings us more beautiful visual effect, makes transportation life more convenient, and improves work efficiency. At present, the processing of envelope image is very critical in the field of classing the envelope image automatically. So, in order to class envelope images more quickly and accurately, the way of processing and presenting envelope image has been a hot topic in the image processing research.In the article, some features of the envelope image have been extracted by using different algorithms. After the features extraction, we encode and analysis the eigenvalue. Firstly, the whole envelope image is processed by using the classical edge detection algorithm—Sobel, and binaried with a const threshold. The window part will be located based on the white pixels density feature. Secondly, in the window area of binary image, some features are extracted by using different algorithms, such as Hough transform and CCA. In the window part of original image, some other features are extracted by using Histogram and GLCM. At last, through analyzing the experimental results, encode the features. At the limited precision, different envelope image owns different coding.The experimental results indicates that the binary threshold feature, the number of connected components, statistic characteristic and so on are more effective than other characteristic in presenting envelope image.