People　are　accustomed　to　seeing　certain　objects　in　specific　and　recurring　contexts　in　daily　life.　Many　research　results　have　indicated　that　consistent　scene　could　result　in　better　object　recognition.　Research　has　also　shown　that　contextual　regularities　and　image　quality　can　help　us　to　form　rapid　judgments　of　particular　objects　and　summon　our　attention　when　the　objects　appear　to　be　incongruent　with　our　expectations.　Based　on　attention　restoration　theory,　this　study　aims　to　investigate　the　influence　of　contexts　on　object　recognition　by　adopting　eye　tracking　techniques.　4　groups　of　college　students　(N=　20)　were　asked　to　view　52　full-color　scene　photographs　with　4　levels　of　image　quality　repeatedly　and　recognize　the　shown　objects　at　the　same　time,　during　which　their　eye　movements　were　recorded.　Results　indicated　that　(1)　compared　with　consistent　object,　participants　recognized　inconsistent　objects　more　slowly　(F(1,28)　=　27.1,　p　＜　0.001),　less　accurately　(F(1,28)　=　19.6,　p　＜　0.001),　fixating　longer　(F(1,28)　=　23.6,　p　＜　0.001),　fixating　more　frequently　(F(1,28)　=　42.3,　p　＜　0.001)　and　more　fixations　prior　(F(1,28)　=　16.9,　p　＜　0.001).　(2)　compared　with　the　objects　with　original　image　quality,　the　objects　with　degradation　qualities　were　harder　to　be　recognized,　presented　less　accurately　(F(1,28)=　8.13,　p＜　0.01),　longer　reaction　time　(F(1,28)=　6.40,　p＜　0.05),　longer　fixation　time　(F(1,28)=　9.68,　p＜　0.01)　and　more　frequent　fixation　(F(1,28)=　10.1,　p＜　0.01).　(3)　the　more　the　image　quality　was　degraded,　the　less　contextual　clues　were　obtained,　which　may　reduce　the　effect　of　semantic　consistency.