It is still a challenge to integrate a biometrics solution such as fingerprint matching into a smart card. However, current generation of smart card is usually equipped with an 8- or 16-bit microcontroller which has limitations on hardware resources. In this paper, we present our approaches to reduce memory requirement for storing image buffer and to reduce the image processing time. Also, we show the advantages of using line extraction scheme to reduce the size of the accumulator array and to speed up the matching process. The proposed method contains more information per bit than the conventional method that can be embedded in smart cards for a real-time match-on-card system implementation. Furthermore, optimization in platform level offers feasibility of code integration into a low memory budget platform. An example of fingerprint identification implementation on a DSP platform with limited memory budget is presented to show effectiveness of our approach.