Design and ASIC implementation of image segmentation algorithm for autonomous MAV navigation

Over the past few years Micro Air Vehicle has gained prominence due to its widespread applications in the field of military and civilian applications. Images captured by onboard cameras on a MAV need to be processed in real time and for this purpose segmentation algorithm is used. On board processing of images is a major challenging task as it involves deciphering images and performing the required operations. In this work, a new image segmentation algorithm has been built using morphological operations, thresholding, edge detection and merging. A detailed analysis has been carried out for choice of appropriate segmentation techniques for design and field-programmable gate array implementation of image segmentation. Software reference model for image segmentation using edge based segmentation and region based segmentation has been developed. The proposed image segmentation algorithm has 17% improvement in Peak Signal-to-Noise Ratio values compared to exiting algorithms. The design consumes power of 0.322W and operates at a maximum frequency of 152.26 MHz in FPGA. Implementation in application-specific integrated circuit occupies less area and power and the working frequency is 188.68 MHz nearly 23.9 percent more than FPGA. The designed algorithm can be extended to include video segmentation and the two color segmentation implementation carried out in this work can also be extended to color image segmentation.