Trapezoid pixel array complementary metal oxide semiconductor image sensor with simplified mapping method for traffic monitoring applications

Abstract

A trapezoid (TZOID) pixel array complementary metal oxide semiconductor (CMOS) image sensor, and a pixel-resolution mapping (P-RM) method to extract information from the observed objects without requiring any specific patterns or man-made markings, are proposed for traffic applications. The size of each pixel in the TZOID imager was determined by pixel design equations based on the relationship between the pixel’s physical size and related spatial resolution. The TZOID imager trades detection accuracy for a tolerable data loss to reduce communication bandwidth and computation power. The P-RM method determines the pixels’ actual resolutions in the real world of the captured imaged by both standard rectangular single- and multiresolution TZOID pixel arrays. Both indoor and outdoor experiments were performed using single- and multiresolution imagers. The TZOID imager integrated circuit (IC) was fabricated in a 0.18-μm CMOS process. It was tested, and its performance parameters were compared with an off-the-shelf CMOS image sensor IC. Tests showed that the TZOID generates 98% less data than that of a standard image sensor, trading off only 2% detection accuracy. Results also show that the proposed method provides a general procedure for designing a multiresolution sensor for traffic applications.