Method and apparatus for optimizing video processing system design using a probabilistic method to fast direct local search

• US 20040010479A1
• Filed: 07/15/2002
• Published: 01/15/2004
• Est. Priority Date: 07/15/2002
• Status: Active Grant

• Current Assignee: Pendragon Wireless LLC (Pendrell Corporation)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Ali, Walid
• Granted Patent: US 6,950,811 B2
• Time in Patent Office: Days
• US Class Current: 706/13
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming
• CPC Class Codes: H04N 5/14 Picture signal circuitry for video frequency region cameras or camera modules comprising electronic image sensors, or control thereof H04N23/00

Abstract

A method and apparatus for obtaining an optimum video processing system design, includes the steps of:(a) performing a hybrid method including: (i) using genetic algorithms to find video sample candidates having a best chromosome for performance and associated fitness values; and (ii) using a probabilistic method to find local optima based on the fitness values from step (a)(i) in a local search scheme. The results of the probabilistic method are then reinserted into genetic algorithms to increase speed and reduce the processing requirements if only genetic algorithms were used. The results obtained from the genetic algorithms using the local optima found by the probabilistic method may then be output to one of a video processing system and video module. Hardware for an apparatus and system to perform same includes a video source input, a processor, a genetic algorithm module, a probabilistic search module, a memory and at least one video module.

Claims

2. A method for obtaining an optimum video processing system design, comprising the steps of:

(a) using genetic algorithms to find video sample candidates having a best chromosome for performance;

(b) storing the best chromosome found in step (a) and an associated fitness value in a memory;

(c) repeating steps (a) and (b) to obtain a plurality of chromosomes for one of (i) a predetermined number of times and (ii) a predetermined fitness value;

(d) using a probabilistic method to find local optima based on the fitness values stored in memory in step (b);

(e) inserting local optima values into the memory; and

(f) performing genetic algorithms using the local optima values obtained in step (e).

1. A method for obtaining an optimum video processing system design, comprising the steps of:

(a) performing a hybrid method comprising;

(i) using genetic algorithms to find video sample candidates having a best chromosome for performance and associated fitness values; and

(ii) using a probabilistic method to find local optima based on the fitness values from step (a) (i) in a local search scheme;

(b) providing results obtained in step (a) (ii) for further processing by genetic algorithms;

(c) providing results obtained in step (b) to one of a video processing system and video module.

3. The method according to claim 2, further comprising providing the values obtained in step (e) to one of a video module and video processing module.

4. The method according to claim 2, wherein the probabilistic method includes building a probability vector according to the following equations:

5. The method according to claim 4, wherein the probability vector includes values of a predetermined number of recent video processing solutions having the highest fitness values.

6. A device for obtaining an optimum video processing system design, comprising:

a video source input;

a processor communicating with the video source input;

a genetic algorithm module and a probabilistic search module for conducting respective genetic algorithms and probabilistic searches;

a memory communicating with the processor and genetic algorithm module and probabilistic search module for storing respective values of chromosomes and associated fitness values;

wherein the probabilistic search module retrieves data from the memory stored by the genetic algorithm module and performs searches to obtain local optima, results of which are stored in the memory for further processing by the genetic algorithm modules.

7. The apparatus according to claim 6, wherein the probabilistic search module performs a search according to the following equations:

6-1. The apparatus according to claim 5, wherein the processor further comprises output means for providing optimum parameters to at least one video module.

8. A system for obtaining optimum video processing design, comprising:

means for performing a hybrid method for optimization by;

(i) using genetic algorithms to find video sample candidates from a video source 302 having a best chromosome for performance and associated fitness values; and

(ii) using a probabilistic method to find local optima based on the fitness values from step (a) (i) in a local search scheme;

means for providing results obtained in step (a) (ii) for further processing by genetic algorithms; and

means for providing results obtained in step (b) to one of a video processing system and video module.

9. A system method for obtaining an optimum video processing system design, comprising:

genetic algorithms means for finding video sample candidates having a best chromosome for performance;

storage means for storing the best chromosome found by the genetic algorithm means, and for storing an associated fitness value in a memory;

means for providing a probabilistic search to find local optima based on the fitness values stored in said storage means);

means for inserting local optima values into the memory; and

said genetic algorithms means using the local optima values obtained by the means for providing a probabilistic search.

10. The system according to claim 9, wherein said genetic algorithm means provides the values obtained using the local optima values obtained by the probabilistic search means to one of a video module and video processing module.

11. The method according to claim 9, wherein the means for providing a probabilistic search builds a probability vector according to the following equations: