Masaaki Matsumura

A study on an evolutionary pixel predictor and its properties

Evolutionary methods based on genetic programming (GP) enable dynamic algorithm generation, and have been successfully applied to many areas such as plant control, robot control, and stock market prediction. However, conventional image/video coding methods such as JPEG and H.264 all use fixed (non-dynamic) algorithms without exception. In this article, we introduce a GP-based image predictor that is specifically evolved for each input image. Experimental results demonstrate 2.9 % less entropy (overhead included) than CALICs gradient adjusted predictor.

Generating subject oriented codec by evolutionary approach

Many image/video codecs are constructed by the combination of various coding tools such as block division/scanning, branch selection and entropy coders. Codec researchers are developing new coding tools, and seeking versatile combinations that offer improved coding efficiency for various images/videos. However, because of the huge amount of the combination, deriving the best combination is impossible by man-power seeking. In this paper, we propose an automatic optimization method for deriving the combination that suits for categorized pictures. We prepare some categorised pictures, and optimize the combination for each category. In the case of optimization for lossless image coding, our method achieves a bit-rate reduction of over 2.8% (maximum) compared to the combination that offers the best bit-rate averagely prepared beforehand.

Automatic pixel predictor construction using an evolutionary method

Evolutionary methods based on genetic programming (GP) enable dynamic algorithm generation, and have been successfully applied to many areas such as plant control, robot control, and stock market prediction. However, conventional image/video coding methods such as JPEG and H.264 all use fixed (non-dynamic) algorithms without exception. In this article, we introduce a GP-based image predictor that is specifically evolved for each input image. Preliminary results demonstrate 1.4 % and 1.7 % entropy reduction (overhead included) against the optimal linear predictor and CALICs gradient adjusted predictor, respectively.

Accelerating pixel predictor evolution using edge-based class separation

Evolutionary methods based on genetic programming (GP) enable dynamic algorithm generation, and have been successfully applied to many areas such as plant control, robot control, and stock market prediction. However, one of the challenges of this approach is its high computational complexity. Conventional image/video coding methods such as JPEG and H.264 all use fixed (non-dynamic) algorithms without exception. However, one of the challenges of this approach is its high computational complexity. In this article, we introduce a GP-based image predictor that is specifically evolved for each input image, as well as local image properties such as edge direction. Via the simulation, proposed method demonstrated ∼180 times faster evolution speed and 0.02–0.1 bit/pel lower bit rate than previous method.