Abstract: An apparatus and method for computing optima of a function of a digital image. The apparatus includes circuitry configured to initialize a plurality of candidate points that lie in a solution space of the function, and computes one or more stationary points of the function. The circuitry deflates a gradient of the function at each of the one or more computed stationary points, and repeats the computing and the deflating until a first criteria is satisfied. The circuitry selects a predetermined number of fit points, recombines the selected fit points to generate a set of new candidate points, and repeats, for the set of new candidate points, the computing, the deflating, the first repeating, the selecting, and the recombining, until a second criteria is satisfied. The circuitry obtains the optima of the function upon the second criteria being satisfied, and processes the digital image based on the obtained optima.

Abstract: An apparatus and method for computing optima of a function of a digital image. The apparatus includes circuitry configured to initialize a plurality of candidate points that lie in a solution space of the function, and computes one or more stationary points of the function. The circuitry deflates a gradient of the function at each of the one or more computed stationary points, and repeats the computing and the deflating until a first criteria is satisfied. The circuitry selects a predetermined number of fit points, recombines the selected fit points to generate a set of new candidate points, and repeats, for the set of new candidate points, the computing, the deflating, the first repeating, the selecting, and the recombining, until a second criteria is satisfied. The circuitry obtains the optima of the function upon the second criteria being satisfied, and processes the digital image based on the obtained optima.