Embedded Image Retrieval System Based On Multi-scale Space Theory

Nowadays, Content Based Image Retrieval (CBIR) is a hotspot in image processing. Early CBIR methods mainly focus on describing the bottom information of images, such as color, texture, shape and so on. These methods have some achievements but they have more difficulties to describe image scaling, rotation movement, affine and other features in detail. Therefore, researching algori thm of image description and its application in multi-scale space has become a promising choice for CBIR. In this paper, according to the current research trends, we mainly have done the work as fallows:(1) With analyzing the current status of research scale invariant feature transform(SIFT) and its related defects, adaptive feature extraction and image matching based on Haar Wavelet Transform and SIFT (AHWT-SIFT) is proposed in this paper. Feature points extracted by our algorithm can satisfy the request of detailed describing the scale and rotation invariant characteristics of images, and our algorithm provide favorable theoretical basis for further research on image recognition, image tracking, and image retrieval.(2) The current paper presents a robust image salient region extraction and matching algorithm based on the maximally stable extremal regions in the difference of Gaussian scale space (DoGSS-MSERs) combined with scale-invariant feature transform (SIFT) algorithm and the maximally stable extremal regions (MSERs) algorithm. The feature regions extracted in the current paper inherit the good properties of SIFT and MSERs (scale-invariant and affine-invariant) and are more stable and more accurate for matching. It makes up for SIFT which has affine-invariance deficiency and for MSERs which have scale-invariance deficiency, and can be used for the identification of the object in the image, image orientation, image retrieval, as well as real-time image processing and other aspects.(3) With the feature points or areas extracted in the paper. Image retrieval algorithm based on multi-scale space theory has been come true. The experiment proved that it is possible to greatly improve the accuracy of the image retrieval with our method. And it provides a strong basis for the further expanding the application of image retrieval algorithm.(4) Combined with the development of multimedia technology and3G communication technology, image retrieval techniques are applied to embedded systems. We have achieved the real-time image retrieval system based on the theory of multi-scale space, which can send out the request from the development board terminal, and the PC-server accepts the request to calculate, retrieval, and sent image retrieval results and so on.The paper is based on the theory of multi-scale to analysis and process image, and then extracts the image description which has a scale-invariant, afl’ine invariant, rotation invariant, illumination invariant characteristics. And it provides a favorable theoretical basis for further research of image recognition, image tracking and image retrieval.