A New Car Seat Detection Method

Abstract

Abstract. In order to detect the products of car seat, propose a new car seat detection method in this paper. First, the car seat image is normalized to 256 pixels *256 pixels image. Then extract SIFT (scale invariant feature transform) feature points and match the points. According to the position of two matched points, the matching results are divided into two categories. One is vertical match point; the other is tilt matching point. Compare the number of the two categories. When the vertical match points are more than the tilt matching points, the answer is correct. Otherwise the answer is wrong. Experimental results show that for the detection of three different types of car seats, the detection accuracy is higher than 98%.