Summary: | 碩士 === 國立勤益科技大學 === 電子工程系 === 101 === With the increasing development of information technology, internet has become one of the modern major means of communication. The Internet auction market is also booming, where costumes are the most frequently purchased items. However, people have experienced that clothes on mannequin is nice, but the actual product received is found not suitable for them. Therefore, if there is virtual fitting system, you can match online to confirm whether it is right to buy so to save users a lot of time and money and so on.
This study takes portrait and jacket images by a CCD camera. By pro-processing images, color images are separated into R, G, B three planes and grayscale processing, a total of four images. Then it is followed by using the Sobel filter to detect the measured edge of portraits and costumes images.
It is also through closing the edge image gaps and filling the edge image to achieve completing capture of portraits and clothes. Using the full portrait histogram analysis again will partition the shoulder line and midline of the portrait. As for the costumes the highest point of neckline is used for positioning. At last through the inspection of portrait shoulder width and upper body length, and scaling shirt size, the jacket can be set on users based on located points.
Experimental results show that using this method can accurately locate the shoulder line and the midline of a jacket, so it is applied to facilitate the user to achieve the effect of fitting.
|