Banding and doubling are two of the most common problems in inkjet printing, which are caused by printheads defects including blocked nozzle and poor registration. It is expensive and inefficient to solve the problems by enhancing device accuracy or replacing the defective printhead, since the device or printhead may degrade over time. In this paper, we propose an economical and efficient method for banding and doubling reduction by detection of printhead defects without requirement of hardware renewal. We design a test pattern composed of specially organized dots to diagnose status of each nozzle by image processing. Specifically, after analyzing the printed and scanned test pattern, halftone images are modified according to the detected information for banding and doubling reduction. We adjust the size and placement of dots based on human perception in order to improve the printing quality. The detection and reduction are carried out in CMYK printing colorant channels separately. Thus, it could overcome the inter-channel interference which may affect the detection accuracy. Experimental results show that the proposed technique is flexible, highly precise and has extensive self-adaptability for various devices, halftone styles, substrates and inks.