This paper discusses the use of image processing technique to detect lizards in a video stream. We discuss the categorization of form perception, human perception, and animal perception. The sub-categories of them are briefly discussed. We defragmented the video into a total of 3459 images, some are with only background scene, some contains lizard. We discuss how we apply background subtraction to segment out the lizard, followed by experiments comparing thresholding values and methods. We achieve an encouraging average hit rate of 98% and average computation time of 1.066 seconds.