An analog color video signal is sampled at a frequency which equals a multiple of a frequency of a chrominance subcarrier. The sampled signal is converted into a corresponding first digital signal. The first digital signal is converted into a second digital signal by changing a data arrangement. The second digital signal is processed into a coded signal through a predictive coding process. During the predictive coding process, first data of the second signal are predicted on the basis of second data of the second signal. The first data correspond to a picture element in a first scanning line. The second data correspond to a picture element in a second scanning line adjacent to the first scanning line. The second data picture element is selected from picture elements in the second scanning line at which phases of the chrominance subcarrier are equal to a phase of the chrominance subcarrier at the first data picture element. The selected picture element is spatially closest to the first data picture element. The predicted first data are used in the predictive coding process.

The present invention aims to execute effective compression even in the case where inputted image information has the bit number smaller than that of a pixel value, such as a tag, and besides in the case where the entropy increases if a prediction error obtained by subtraction of a value of a neighboring pixel from that of an object pixel is used for calculation of the entropy. According to the present invention, in an image encoding apparatus, the inputted image information is made to be packed image data of a fixed length and provided to a couple of predicting elements and a couple of prediction error calculating elements. The packed image data and prediction errors outputted from the couple of prediction error calculating elements are provided to a prediction error selecting element and one of the data and prediction errors having a smallest entropy is selected. A selecting element outputs an identification number of one of the predicting elements which succeeds in prediction, and if the predicting elements fail in prediction, the selecting element outputs the data or prediction error selected by the prediction error selecting element. The output from the selecting element is encoded by an encoding element to Huffman code, for example

In a predictive encoder device supplied with an input image signal divisible into odd and even number line signals, first and second predictive encoders (32, 37) individually carry out predictive encoding of the odd and the even number line signals into first and second encoded signals at a comparatively low speed by the use of first and second local decoded signals, respectively. The first and the second local decoded signals are supplied to the second and the first predictive encoders through delay circuits (42, 43), respectively, to predict a current picture element with reference to picture elements located on adjacent lines. Two delay circuits (36, 44) are connected to an output terminal (32c) of the first predictive encoder and to an input terminal (37a) of the second predictive encoder to adjust timings of the first and the second encoded signals. A predictive decoder device also comprises first and second predictive decoders (62, 83) corresponding to the first and the second predictive encoders.