Variable block based motion estimation using hexagon diamond full search algorithm (HDFSA) via block subtraction technique

• Main Author: Hardev Singh, Jitvinder Dev Singh
• Format: Thesis
• Language: English; English
• Published: 2015
• Subjects: T Technology (General); TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utem.edu.my/id/eprint/16853/1/Variable%20Block%20Based%20Motion%20Estimation%20Using%20Hexagon%20Diamond%20Full%20Search%20Algorithm%20%28HDFSA%29%20Via%20Block%20Subtraction%20Technique.pdf; http://eprints.utem.edu.my/id/eprint/16853/2/Variable%20block%20based%20motion%20estimation%20using%20hexagon%20diamond%20full%20search%20algorithm%20%28HDFSA%29%20via%20block%20subtraction%20technique.pdf; http://eprints.utem.edu.my/id/eprint/16853/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=96175

Motion estimation is a technique to reduce high information redundancy which exists between successive frames in a video sequences. There are many types of motion estimation method but the most used method is the block matching method which is the fixed block matching and the variable block matching. The fixed block matching uses the same block size throughout the motion estimation process while the variable block matching uses different block size. The variable block matching developed based on four stages which is the video and frame selection, threshold calculation, block size selection and search pattern. In the video and frame selection, pre-defined video which have different type of motion and size is used for the algorithm evaluation purpose. The threshold calculation is based on the video selected. Each video selected will have its own threshold which is used for the block size selection. There is three block size selection which is 16×16 pixels block size (uniform motion), 8×8 pixels block size (moderate motion) and 4×4 pixels block size (complex motion). In order to calculate the threshold and block size selection, the block subtraction technique is implemented. The concept of the block subtraction technique is based on the changes of pixels value between successive frames which represent the existence of motion. The next stage of algorithm development is the search pattern which is the hexagon diamond (16×16 and 8×8 pixels block size) and full search pattern (4×4 pixels block size). To evaluate the performance of the developed algorithm, the average PSNR value, average search point and average elapsed processing time is calculated. Overall, the developed algorithms have similar PSNR value and lower average search point compared to superior algorithms. The average elapsed processing time have increased due to the implementation of the block subtraction technique and the variable block matching.