Ultrasonic　imaging　technique　can　provide　more　intuitive　and　reliable　detection　results　for　detection　personnel,　and　is　widely　used　in　ultrasonic　nondestructive　testing.　However,　the　ultrasonic　detection　signal　has　wide　time　width　and　the　time-resolution　is　low.　When　two　defects　are　adjacent　in　distance,　the　echoes　from　the　two　defects　are　overlapped;　it　is　difficult　to　identify　the　adjacent　defects　in　ultrasonic　imaging.　In　order　to　increase　the　time　resolution　of　ultrasonic　detection　signal　and　achieve　adjacent　defect　identification,　a　pulse　compression　technique　was　developed　and　applied　in　ultrasonic　array　imaging,　which　combines　Wiener　filtering　and　autoregressive　spectral　analysis　techniques,　and　the　identification　of　adjacent　defects　can　be　achieved.　The　influence　of　the　autoregressive　order　and　the　attenuation　window　width　on　defect　identification　result　in　autoregressive　spectral　analysis　was　studied;　and　the　optimum　parameters　were　obtained.　Simulation　and　detection　experiment　results　show　that　the　proposed　technique　can　greatly　improve　the　time　resolution　of　the　ultrasonic　detection　signal　and　achieve　adjacent　defect　identification.