The title of this study was Retrieval Citra Batik Pesisir. The purpose of this study is how to measure the similarity between the query image to the image database on the image of a very diverse coastal batik influenced by the geographical location of the islands, the state of nature, and the structure of society, by implementing the transformation method curvelet fast discrete and HSV color space. And knowing the average yield precision image retrieval based coastal batik content using relative method manhatan. This research uses experimental analysis on coastal batik image dataset of the query image with the image in the database. Coastal batik image data taken are secondary data from the internet already results reproduk regardless of noise and shooting techniques. Applications are designed and built in stages as a test prototype image retrieval coastal batik using programming language Matlab R2014a. Based on the description of explanation and discussion on the overall chapter thesis for image retrieval batik to determine the extraction of the best features that the average value of precision curvelet scale of 4 to 85.17%, curvelet scale of 5 namely 88.62%, curvelet scale of 6 average precision value ie 90.73 %. For the average value of precision for HSV color space by 81.55%. As for the average value of precision curvelet scale 4 and HSV color space by 83.44%, to curvelet scale of 5 and HSV color space by 85.34% and to curvelet scale 6 and HSV color space by 87.71%. From this it can be deduced that the higher the measurement scale, the average precision is getting better at the time of image retrieval.Keywords: Content Based Image retrieval, Transformasi Curvelet Fast Discrete , Ruang Warna HSV , Relative Manhatan, Precision.