A high redshift quasar is useful to investigate the early part of our universe. Since they are one of the brightest objects in the early universe, they can provide us with clues of the growth of super massive black holes and the early metal enrichment history. To discover the high redshift quasars, we designed a survey of wide area and moderate depth; Infrared Medium-deep Survey (IMS), a J-band imaging survey of ~200 $deg^2$ area where the multi-wavelength data sets exist. To obtain the J-band data, we are using the United Kingdom Infra-Red Telescope (UKIRT), and so far we have covered ~20 $deg^2$ with Y- or J-bands over three observing runs during 2009. We used color-color diagrams of multi-wavelength bands including i, z, Y, J, K, $3.6{mu}m$ and $4.5{mu}m$ to select high redshift quasars. The major challenge in the selection is many M/L/T dwarfs, low redshift galaxies, and instrumental defects that can be mistaken as a high redshift quasar. We describe how such contaminating sources can be excluded by adopting multiple color-color diagrams and eye-ball inspections. So far, our selection reveals two quasar candidates at z~7.