Abd Majid Darsono

A new Orthogonal - Diamond Search algorithm for Motion Estimation

Many fast Block Matching Algorithm or BMA have been proposed and developed in recent years in order to reduce the computational cost while maintaining the quality of the video signal. In this paper, a new Orthogonal-Diamond Search (ODS) is proposed which employs an orthogonal-shaped search pattern in the first step and then is switched into small diamond-shaped search pattern for the next step. The proposed algorithm and several established algorithms, namely Full Search (FS), Three-Step Search (TSS) and Diamond Search (DS) are implemented using MATLAB and their performance are being compared and analyzed in terms of computational complexity, peak signal-to-noise ratio (PSNR), and number of search points. Simulation result shows that the proposed algorithm can find motion vector with fewer number of search points while maintains close performance of video quality with other selected algorithms.

Vector-based technique for distance computation in narrow phase collision detection

Distance computation between three-dimensional (3D) for narrow phase collision detection is always a common problem for researchers to produce high quality optimization with reasonable speed of intersection experiment. Given a set of objects with hundreds to thousands of polygons, finding the nearly exact distance between 3D objects are rather expensive and exhausted. In this paper, we presented our current work on performing distance computation between 3D objects based on vector-based searching using Dynamic Origin Point (DyOP). Our theoretical work suggested that by implementing DyOP, we could reduce the numbers of testing that will be undergone by the simulation and thus avoiding expensive computational cost.

Adaptive Diamond Orthogonal Search Algorithm for Motion Estimation

Implementation of the Block Matching Algorithm (BMA) in a series of video codecs is due to its simplicity and high compression efficiency. Due to this, many fast search methods of BMAs are being introduced to increase the efficiency of the Motion Estimation process. This paper proposed a new algorithm, namely Adaptive Diamond Orthogonal Search Algorithm (ADOS) which employs three different search pattern for its two main stages.. At initial step, an additional step is added to predetermine static block to further speed up the search process as it is beneficial to small motion video sequence contents. The performances of the ADSS are being compared with three selected established algorithms, namely Full Search (FS), Diamond Search (DS) and Hexagon-Diamond Search (HDS). Based on the simulation result, the proposed algorithm yields fewer search points while maintaining close video quality performance with other algorithms.

Computing penetration depth of collision detection between primitives

DyOP-PD is the third part of narrow phase collision detection that involved finding the penetration depth using Dynamic Pivot Point (DyOP) technique. After the object or the primitive has collided into another primitive, penetration depth algorithm will check for any significant penetration that might occurs upon the intersection. In our implementation, we propose our own algorithm called DyOP-PD where the implementation will be done in 2D for Lin Canny, GJK and DyOP technique and 3D for Lin Canny and DyOP technique. In this chapter, we will explain the concept of our DyOP-PD algorithm that has efficiently contributed to faster penetration depth computation when using DyOP technique.

Closest features determination between range point of contact

Contact determination is the second part of narrow phase collision detection that involved finding the point of contact using trajectory information or other technique to approximate time of collision and coordinate the collision. Each object has their own trajectory information that is based on their object direction that change when updating their coordinate of the object vertices. For example, when one of the object vertex located at point A (let say 1,1) move into the point B (let say 1,5), then we have an update just for y-axis where it changes from point 1 to point 5 and involved y-positive direction. Hence, our purpose is to approximate accurately the point B where we need to know the direction of x-axis, y-axis, and z-axis are heading under reasonable time consumption and less computation with nearly accurate approximation technique. In this paper, we presented our theoretical framework of finding closest features from point of contacts between nearly intersected triangles.