Implementation and Validation of Software for Demonstration of New Sensor Features

Abstract

A camera is an optical instrument which is used for recording images, which may be still photographs or sequence of images consisting of video or movies. A camera is considered of excellent quality if it has high resolution which can be increased by increasing the size of pixel which gives more clearer object. The CPD(Consumer Product Design) department of ST Microelectronics, deals with the resolution and object detection of an image. Code is developed in either C or Python to demonstrate the features of the latest sensor under development like face or object detection. Face detection algorithm of STV0987 co processor is used for detecting the face.The algorithm uses integral images for detecting the face over a particular representation of it, called integral image. The integral image is build from the original one assigning to each pixel the value corresponding to the sum of all the pixels lying above and to the left of that pixel in the original image.Every feature is identified by a five element tupple where the first number identify the family which the feature belongs, the second and the third indicate the row and the column of the upper-left corner of the feature while the fourth and the fifth indicate the width and height of the smallest sub-rectangle of a feature in pixel. MVSS (Machine Vision Subsystem) is a upgraded version of STV0987 co-processor with upgraded object detection features. It has input resolution of VGA and can detect face size upto 20 pixels with no restriction of range of rotation in plane. A firmware is designed to detect objects from images whose values are extracted in python. Face detection algorithm has both hardware and software partition. Software values that are extracted in python and the hardware values of the module are compared and resolution is compared and hence improved.