Adaptive segmentation method of currency image based on texture features

In this article, we establish a fast and convenient image retrieval system. The currency image is presented through segmentation — the color image is converted to a gray image and then the image is used to create a Haar Discrete Wavelet Transform (HDWT). The next step is to extract the texture feather, adopt the Principal Component Analysis to determine the most suitable weights of each trait and conduct the binary operation. This is followed by a repair using image processing technology to achieve separation of the currency and the background. Image segmentation was conducted on images from different eras, different shapes, and different colors. A comparison with the established image segmentation methods, such as Otsu's and Level Set, as stated in this article, show that the proposed method can overcome the problems regarding imbalances in image decoration and color as well as irregularities; the segmentation can be successfully achieved even though the color of the image background is similar to that on the edge of currency objects.