Video camera having an adaptive automatic iris control circuit

A video camera comprising:

• optical means for gathering incident light from an object to form an image on a predetermined image formation plane,
• incident light amount regulating means, responsive to an incident light amount regulating signal, for regulating an incident light amount;
• image sensing means for image sensing the image of the object within a predetermined frame formed by said optical means to provide a luminance signal;
• luminance distribution characteristics extraction means, connected to said image sensing means, for dividing said predetermined frame into a predetermined plurality of sub-areas and for extracting a luminance distribution of the image in said predetermined frame as a luminance value for each of said sub-areas according to said luminance signal to provide the extracted luminance distribution as a plurality of luminance distribution signals;
• means for generating a target value signal for determining a target value of said incident light amount regulating signal that can be arbitrarily set by a user according to the luminance distribution of the image;
• adaptive means, coupled to said luminance distribution characteristics extraction means and having an artificial neural network to which said plurality of luminance distribution signals are input, for generating said incident light amount regulating signal in accordance with a predetermined conversion, the predetermined conversion being made adaptive so that an offset between a teacher signal and said incident amount regulating signal is minimized, and
• manually operable selecting means, responsive to operation by the user, for selecting one of said first value signal and said incident light amount regulating signal as said teacher signal,
• said artificial neural network of said adaptive means including
• an intermediate layer including a plurality of first converting means, each for obtaining one output by performing a first predetermined conversion of said luminance distribution signals,
• an output layer having a plurality of second converting means, each for obtaining one output by performing a second predetermined conversion of an output of said intermediate layer,
• output converting means for performing a third predetermined conversion of an output of said output layer to generate said incident light amount regulating signal, and
• conversion updating means for adaptively updating said first and second predetermined conversions according to the offset between said incident light amount regulating signal output of said artificial neural network and said teacher signal so that the offset is minimized, each of said first and second converting means including
• weighting means for performing a predetermined weighting of each respective input signal,
• adding means for adding all the weighted signals output by said weighting means, and
• function converting means for converting an output of said adding means according to a predetermined monotone increasing function,
• said conversion updating means comprising weight updating means for updating a weight of each of said weighting means, according to an offer between said incident light amount regulating signal and said teacher signal, by recalculating each weight so that a root mean square of the offset is minimized.

A video camera comprising:

• optical means for gathering incident light from an object to form an image on a predetermined image formation plane;
• incident light amount regulating means, responsive to an incident light amount regulating signal, for regulating an incident light amount;
• image sensing means for image sensing the image of the object within a predetermined frame formed by said optical means to provide a luminance signal;
• luminance distribution characteristics extraction means, connected to said image sensing means, for dividing said predetermined frame into a predetermined plurality of sub-areas and for extracting a luminance distribution of the image in said predetermined frame as a luminance value for each of said sub-areas according to said luminance signal to provide the extracted luminance distribution as a plurality of luminance distribution signals;
• means for generating a target value signal for determining a target value of said incident light amount regulating signal that can be arbitrarily set by a user according to the luminance distribution of the image;
• adaptive means, coupled to said luminance distribution characteristics extraction means and having an artificial neural network to which said plurality of luminance distribution signals are input, for generating said incident light amount regulating signal in accordance with a predetermined conversion, the predetermined conversion being made adaptive so that an offset between a teacher signal and said incident light amount regulating signal is minimized;
• and
• manually operable selecting means, responsive to operation by the user, for selecting one of said target value signal and said incident light amount regulating signal as said teacher signal,
• said means for generating a target value signal including
• manually operable offset generating means for generating an arbitrary offset for the output of said artificial neural network, and
• adding means for generating said target value signal by adding said arbitrary offset provided from said manually operable offset generating means with an output of said artificial neural network.

A video camera comprising:

• image sensing means for generating luminance signals in accordance with light from an object to be imaged;
• light regulating means for regulating an amount of the light from the object incident upon said image sensing means in accordance with a light regulation signal;
• light regulating control means, coupled to said image sensing means, for generating the light regulation signal, said light regulating control means comprising
• average luminance signal means, coupled to said image sensing means, for dividing the luminance signals corresponding to the object into sub-areas and for generating average luminance signals for each of the sub-areas,
• adaptive network means, coupled to said average luminance signal means and including an artificial neural network, for adaptively generating an optimum aperture value signal which is indicative of an optimum aperture of said light regulating means in accordance with a teacher signal indicative of a desired aperture of said light regulating means and the average luminance signals, and
• adder means, coupled to said adaptive network means, for adding an offset to the optimum aperture value signal to generate the light regulating signal, the light regulating signal being supplied to said adaptive network means as the teacher signal;
• and
• manually operable aperture setting means, responsive to selection by an operator of the video camera and coupled to said adder means, for generating a target value signal corresponding to a desired aperture of said light regulating means, the target value signal being input to said adder means as the offset.

A method of generating video image signals using a video camera comprising the steps of:

• generating luminance signals in accordance with light from an object to be imaged;
• regulating an amount of light to be imaged during said step of generating luminance signals in accordance with a light regulating signal;
• and
• generating the light regulation signal, to control said step of regulating an amount of light to be imaged, comprising the substeps of
• dividing the luminance signals corresponding to the object into sub-areas,
• generating average luminance signals for each of the sub-areas,
• adaptively generating an optimum aperture value signal which is indicative of an optimum aperture value of the video camera in accordance with a teacher signal indicative of a desired aperture of the video camera and the average luminance signals,
• adding an offset to the optimum aperture value signal to generate the light regulation signal, the light regulation signal being used as the teacher signal during said substep of adaptively generating an optimum aperture value signal, and
• generating a target value signal corresponding to a desired aperture of the video camera upon manual selection by an operator of the video camera, the target value signal being used as the offset during said substep of adding an offset to the optimum aperture value signal.