Eastman-kodak Company

Yusheng T Tsai

Kenneth Alan Parulski

Majid Rabbani

Lawrence J Bernstein

A method and apparatus is disclosed for extending the dynamic range of an electronic imaging system to permit the use of lower cost, lower resolution devices to digitize and process the analog image signal generated by an electronic image sensor. Multiple images captured at different exposure levels are stored in a framestore. An electronic data processing unit then compares the pixels from the image captured at a normal exposure level to preselected exposure ranges to determine if the pixel is underexposed or overexposed. If the pixel is underexposed, the pixel is replaced with a corresponding pixel from an image captured at a higher exposure level. If the pixel is overexposed, the pixel is replaced with a corresponding pixel from an image captured at a lower exposure level. A two level optimization routine is employed, wherein median exposure valves are used to compensate for subject motion.

Method and apparatus for extending the dynamic range of an electronic

The apparatus and method of the present invention operates upon the problem of contouring that is encountered in the digital image processing field. The preferred apparatus embodiment of the invention forms quantizing steps that vary in size during the processing of signals in order to minimize contouring effects. The embodiment includes a transforming function block for receiving digital input signals each representing a magnitude value requiring quantization, the transforming function block provides an output gain value that is a function of the magnitude of the value of each received digital input signal. A quantizer is provided for receiving the digital input signals and for performing quantizing steps on the digital input signals to provide a multibit quantized digital signal. A first randomizer is used to randomize the least significant bits of the multibit quantized digital signal when the output gain value is less than or equal to threshold value. Lastly there is provided a second randomizing means for randomizing a number of bits of the multibit quantized digital signal when the output gain value from the transforming function block is greater than the threshold value.

Apparatus and method for dynamic step quantization

The method of the present invention operates upon the imaging signals from a color filter array image sensor to compress the image signals prior to color interpolation by separating the signals into three color planes. Missing green pixels are recovered through interpolation methods, either by bilinear or other techniques, in order to form color ratio signals long B/G.sub.missing and log R/G.sub.missing which allows for proper interpolation after decompression without color artifacts. Upon playback or reception, an inverse compression is applied and missing green, red and blue pixels are interpolated to provide three full planes of color image data.

Compression method and apparatus for single-sensor color imaging systems

The method of the present invention operates upon the imaging signals from a color filter array image sensor to compress the image signals prior to color interpolation by separating the signals into three color planes. Missing green pixels are recovered through interpolation methods, either by bilinear or other techniques, in order to form color ratio signals log B/G.sub.missing and log R/G.sub.missing which allows for proper interpolation after decompression without color artifacts. Upon playback or reception, an inverse compression is applied and missing green, red and blue pixels are interpolated to provide three full planes of color image data.

In a method for creating a scan sequence for a single chip color camera an analysis of typical images, captured through a color filter array, is performed to determine an optimal scan sequence for the particular array and/or the typical images. The images, after being digitally captured, are separated into red, green and a blue image databases and the color databases are processed separately. A Discrete Cosine Transform is performed on blocks of the image data and the average RMS values for the coefficients of the images are determined. The RMS values are sorted in descending order to produce a descending scan sequence that optimizes the performance of run length coding schemes. The scan sequence can be stored in a hardware, firmware or software lookup table as a list of block coordinates or indices and used by the camera system to convert two dimensional blocks of coefficients into one dimensional lists of coefficients suitable for run length coding. The block coefficients are used to convert the decoded coefficients into image blocks before presentation on a color CRT of before producing a color print. By incorporating the method into an image capture system an adaptive system which will optimize coding for different image environments and/or different color filter arrays suitable for the different environments is produced.

Digital compression method and system with improved coding efficiency

The system and method of the present invention compresses digitized color image signals in real time with an error tolerant compression algorithm utilizing block truncation coding. The system operates upon the digitized color components Y, I and Q derived from scanning an image plane. Memory buffers, under control of an address generator, receive digitized color components and store them in a 4.times.4 block format. The system further subdivides each Y and I block into four 2.times.2 blocks. An I.sub.avg. signal is derived by averaging 4 neighboring pixels in each 2.times.2 block of the I image plane, and a Q.sub.avg. signal is derived by averaging the 16 pixel values forming the 4.times.4 block of the Q image plane. Signal compression is accomplished by representing the Y and I blocks with two values and a bitmap. Three pipelined signal channels work in parallel to real-time process the signals. The processed signals are encoded for transmission over a channel. At the channels receiver a decoder decodes the signals and a decompression circuit reconstructs the Y, I and Q signals.

Growing community of inventors

Rochester, NY, United States of America

Yusheng T Tsai