Method and apparatus for scene change detection in digital video compression
AssigneeInternational Business Machines Corporation
InventorsCharlene Ann Gebler, Barbara Ann Hall, Agnes Yee Ngai
Life of this patent
Estimating the term…
Abstract
A method of encoding a picture in an MPEG2 compliant digital video encoder. The method calculates a contrast function, Contrast=.SIGMA..vertline.P(j)-P(j+1).vertline. and thereafter calculates a quantization adjustment function therefrom M(i+1)=C(i+1)/C(i)!M(i), where C=Contrast, P(j) is the luminance or chrominance of the j.sup.th pixel, and M(i)is the average quantization of the i.sup.th picture. The quantization or picture type is adjusted in response to the contrast function, C.
Claims3
In a method of encoding a digital video bitstream in a digital video encoder to produce a compressed digital a video bitstream, said method comprising:
- a) performing intra-picture video compression comprising the steps of:
- a1) discrete cosine transforming a digital image of pixels in a macroblock of a picture to convert the picture from a time domain digital image of the pixels in the macroblock of the picture to a frequency domain image of the pixels in the macroblock;
- a2) quantizing the frequency domain image of the pixels in the macroblock;
- a3) run length coding the quantized frequency domain image of the pixels in the macroblock;
- and
- a4) Huffman coding the run length coded frequency domain image of the pixels in the macroblock;
- and
- b) performing inter-picture compression between a current intra-picture compressed picture and a reference intra-picture compressed picture comprising the steps of:
- b1) inverse quantizing macroblocks of the current picture;
- b2) inverse discrete cosine transforming inverse quantized macroblocks of the current picture;
- b3) searching from a macroblock in the reference picture to the current picture for a nearby macroblock in the current picture with same content to the macroblock in the reference picture;
- b4) extracting a chrominance and a luminance difference and a vector between pixels in the current picture macroblock and the reference picture macroblock;
- and
- b5) compressing the chrominance and the luminance difference and the vector between pixels in the current picture macroblock and the reference picture macroblock for transmission;
- the improvement comprising determining a difference in complexity between the current picture and the reference picture and adjusting the quantization therebetween in response to the difference in complexity wherein the difference in complexity is determined by:
- 1) determining the chrominance or luminance, P(j), of individual pixels, where P(j) is the luminance or chrominance of a j.sup.th pixel;
- 2) calculating a contrast function that is the sum over the pixels in a picture of the contrast difference between sets of adjacent pixels, given by
- C=.SIGMA..vertline.P(j)-P(j+1).vertline.,
- where C=Contrast, where large value of C indicates more complexity and details in the picture then a small value of C;
- 3) calculating therefrom a quantization adjustment
- M(i+1)={C(i+1)/C(i)}M(i),
- where M(i) is the average quantization factor used to quantize discrete cosine transformation coefficients of the i.sup.th picture;
- and
- 4) adjusting the quantization or picture type responsive to the ratio of the average quantization factors, so that when encoding goes from a picture of small C to a next picture of larger C, a relatively larger quantization is provided to maintain a relatively equal number of bits produced in each picture.
Dependencies are inferred from the claim text.