Abstract: This paper proposes the design and implementation of object counting algorithm based on image processing by using Raspberry pi on real time basis. Features of an image containing objects to be detected are extracted uses a method called BLOB (Binary Large Object) analysis. This method is used so as to identify the objects/regions called BLOBs using pixel connectivity and further more parameters like area, centroid, and number of BLOBs available in the test image. Preprocessing of the input test image is done prior to the BLOB analysis like thresholding to get a binary image. The number of objects identified is displayed or read out using a speaker interfaced to the hardware. The hardware Raspberry Pi, running with UBUNTU operating system flashed into a memory card, works as a standalone device with the Raspberry Pi camera being used as an image acquisition device to constantly capture image frames of objects of interest.
Keywords: BLOB analysis, object counting, Raspberry Pi.