This paper presents vision-based helipad detection algorithm and implementation of embedded landing system for an UAV (Unmanned Aerial Vehicle). The helipad detection algorithm includes various image processing processes and feature recognition methods. It is essential for the embedded system that the detection algorithm should be simple and accurate. We propose optimized image processing algorithm and simple morphological algorithm to implement an embedded system for detecting the international standard helipad. First, we perform image preprocessing includes filtering, thresholding and labeling. After image preprocessing, we perform feature extraction of the helipad. Second, the size and position calculation of the detected helipad is performed. Third, the size and position information of helipad is transmitted via communication port. In these processes, we can estimate the position information of the UAV by using the size and the position of helipad in the aerial moving image. In experiment, we use aerial moving image photographed by coaxial rotor helicopter and save the detected helipad to SD-card to demonstrate the accuracy and real time processing capabilities. The experimental results show that proposed embedded system has features of real time and reliable accuracy.