ACCURATE AND FAST DYNAMIC SUBPIXEL BASED TUMOR IMAGE RECOGNITION

In recent years, the image processing technique has been applied in many significant fields, such as military, aerospace, and health industry. With the development of medical treatment, the computer-assisted diagnosis and treatment system play increasingly important roles. Methods: This paper presents a new fast and accurate image matching algorithm, which uses Breast Cancer B-scan images as test images. We first present a novel method to obtain the integer-pixel matching result, namely district-identification method. Then, we introduce gradient algorithm to match the sub-pixel position. Results: Experimental results show that the proposed method is about 2.3 times faster than traditional methods and can be suitable for images of arbitrary size. Moreover, it can improve performance in sub-pixel level and the recognition rate is above 96.30%. Conclusion: This paper proposes a novel image matching method to help medical experts to recognize the tumor location fast and accurately, find out the radiotherapy and chemotherapy target, and make full use of the medicines.